2 Format: { eager | lazy }
4 By default, unaccepted memory is accepted lazily to
5 avoid prolonged boot times. The lazy option will add
6 some runtime overhead until all memory is eventually
7 accepted. In most cases the overhead is negligible.
8 For some workloads or for debugging purposes
9 accept_memory=eager can be used to accept all memory
12 acpi= [HW,ACPI,X86,ARM64,RISCV64,EARLY]
13 Advanced Configuration and Power Interface
14 Format: { force | on | off | strict | noirq | rsdt |
16 force -- enable ACPI if default was off
17 on -- enable ACPI but allow fallback to DT [arm64,riscv64]
18 off -- disable ACPI if default was on
19 noirq -- do not use ACPI for IRQ routing
20 strict -- Be less tolerant of platforms that are not
21 strictly ACPI specification compliant.
22 rsdt -- prefer RSDT over (default) XSDT
23 copy_dsdt -- copy DSDT to memory
24 For ARM64 and RISCV64, ONLY "acpi=off", "acpi=on" or
25 "acpi=force" are available
27 See also Documentation/power/runtime_pm.rst, pci=noacpi
29 acpi_apic_instance= [ACPI,IOAPIC,EARLY]
31 2: use 2nd APIC table, if available
32 1,0: use 1st APIC table
35 acpi_backlight= [HW,ACPI]
36 { vendor | video | native | none }
37 If set to vendor, prefer vendor-specific driver
38 (e.g. thinkpad_acpi, sony_acpi, etc.) instead
39 of the ACPI video.ko driver.
40 If set to video, use the ACPI video.ko driver.
41 If set to native, use the device's native backlight mode.
42 If set to none, disable the ACPI backlight interface.
44 acpi_force_32bit_fadt_addr [ACPI,EARLY]
45 force FADT to use 32 bit addresses rather than the
46 64 bit X_* addresses. Some firmware have broken 64
47 bit addresses for force ACPI ignore these and use
48 the older legacy 32 bit addresses.
50 acpica_no_return_repair [HW, ACPI]
51 Disable AML predefined validation mechanism
52 This mechanism can repair the evaluation result to make
53 the return objects more ACPI specification compliant.
54 This option is useful for developers to identify the
55 root cause of an AML interpreter issue when the issue
56 has something to do with the repair mechanism.
58 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG]
59 acpi.debug_level= [HW,ACPI,ACPI_DEBUG]
61 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI
62 debug output. Bits in debug_layer correspond to a
63 _COMPONENT in an ACPI source file, e.g.,
64 #define _COMPONENT ACPI_EVENTS
65 Bits in debug_level correspond to a level in
66 ACPI_DEBUG_PRINT statements, e.g.,
67 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ...
68 The debug_level mask defaults to "info". See
69 Documentation/firmware-guide/acpi/debug.rst for more information about
70 debug layers and levels.
72 Enable processor driver info messages:
73 acpi.debug_layer=0x20000000
74 Enable AML "Debug" output, i.e., stores to the Debug
75 object while interpreting AML:
76 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
77 Enable all messages related to ACPI hardware:
78 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
80 Some values produce so much output that the system is
81 unusable. The "log_buf_len" parameter may be useful
82 if you need to capture more output.
84 acpi_enforce_resources= [ACPI]
86 Check for resource conflicts between native drivers
87 and ACPI OperationRegions (SystemIO and SystemMemory
88 only). IO ports and memory declared in ACPI might be
89 used by the ACPI subsystem in arbitrary AML code and
90 can interfere with legacy drivers.
91 strict (default): access to resources claimed by ACPI
92 is denied; legacy drivers trying to access reserved
93 resources will fail to bind to device using them.
94 lax: access to resources claimed by ACPI is allowed;
95 legacy drivers trying to access reserved resources
96 will bind successfully but a warning message is logged.
97 no: ACPI OperationRegions are not marked as reserved,
98 no further checks are performed.
100 acpi_force_table_verification [HW,ACPI,EARLY]
101 Enable table checksum verification during early stage.
102 By default, this is disabled due to x86 early mapping
105 acpi_irq_balance [HW,ACPI]
106 ACPI will balance active IRQs
109 acpi_irq_nobalance [HW,ACPI]
110 ACPI will not move active IRQs (default)
113 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
114 Format: <irq>,<irq>...
116 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
118 Format: <irq>,<irq>...
120 acpi_mask_gpe= [HW,ACPI]
121 Due to the existence of _Lxx/_Exx, some GPEs triggered
122 by unsupported hardware/firmware features can result in
123 GPE floodings that cannot be automatically disabled by
125 This facility can be used to prevent such uncontrolled
127 Format: <byte> or <bitmap-list>
129 acpi_no_auto_serialize [HW,ACPI]
130 Disable auto-serialization of AML methods
131 AML control methods that contain the opcodes to create
132 named objects will be marked as "Serialized" by the
133 auto-serialization feature.
134 This feature is enabled by default.
135 This option allows to turn off the feature.
137 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
140 acpi_no_static_ssdt [HW,ACPI,EARLY]
141 Disable installation of static SSDTs at early boot time
142 By default, SSDTs contained in the RSDT/XSDT will be
143 installed automatically and they will appear under
144 /sys/firmware/acpi/tables.
145 This option turns off this feature.
146 Note that specifying this option does not affect
147 dynamic table installation which will install SSDT
148 tables to /sys/firmware/acpi/tables/dynamic.
150 acpi_no_watchdog [HW,ACPI,WDT]
151 Ignore the ACPI-based watchdog interface (WDAT) and let
152 a native driver control the watchdog device instead.
154 acpi_rsdp= [ACPI,EFI,KEXEC,EARLY]
155 Pass the RSDP address to the kernel, mostly used
156 on machines running EFI runtime service to boot the
157 second kernel for kdump.
159 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
160 Format: To spoof as Windows 98: ="Microsoft Windows"
162 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
163 of 2 which is mandated by ACPI 6) as the supported ACPI
164 specification revision (when using this switch, it may
165 be necessary to carry out a cold reboot _twice_ in a
166 row to make it take effect on the platform firmware).
168 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
169 acpi_osi="string1" # add string1
170 acpi_osi="!string2" # remove string2
171 acpi_osi=!* # remove all strings
172 acpi_osi=! # disable all built-in OS vendor
174 acpi_osi=!! # enable all built-in OS vendor
176 acpi_osi= # disable all strings
178 'acpi_osi=!' can be used in combination with single or
179 multiple 'acpi_osi="string1"' to support specific OS
180 vendor string(s). Note that such command can only
181 affect the default state of the OS vendor strings, thus
182 it cannot affect the default state of the feature group
183 strings and the current state of the OS vendor strings,
184 specifying it multiple times through kernel command line
185 is meaningless. This command is useful when one do not
186 care about the state of the feature group strings which
187 should be controlled by the OSPM.
189 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
190 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
191 can make '_OSI("Windows 2000")' TRUE.
193 'acpi_osi=' cannot be used in combination with other
194 'acpi_osi=' command lines, the _OSI method will not
195 exist in the ACPI namespace. NOTE that such command can
196 only affect the _OSI support state, thus specifying it
197 multiple times through kernel command line is also
200 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
203 'acpi_osi=!*' can be used in combination with single or
204 multiple 'acpi_osi="string1"' to support specific
205 string(s). Note that such command can affect the
206 current state of both the OS vendor strings and the
207 feature group strings, thus specifying it multiple times
208 through kernel command line is meaningful. But it may
209 still not able to affect the final state of a string if
210 there are quirks related to this string. This command
211 is useful when one want to control the state of the
212 feature group strings to debug BIOS issues related to
215 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
216 '_OSI("Module Device")' FALSE.
217 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
218 '_OSI("Module Device")' TRUE.
219 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
221 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
223 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
224 they all will make '_OSI("Windows 2000")' TRUE.
227 Override the pmtimer bug detection: force the kernel
228 to assume that this machine's pmtimer latches its value
229 and always returns good values.
231 acpi_sci= [HW,ACPI,EARLY] ACPI System Control Interrupt trigger mode
232 Format: { level | edge | high | low }
234 acpi_skip_timer_override [HW,ACPI,EARLY]
235 Recognize and ignore IRQ0/pin2 Interrupt Override.
236 For broken nForce2 BIOS resulting in XT-PIC timer.
238 acpi_sleep= [HW,ACPI] Sleep options
239 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig,
240 s4_nohwsig, old_ordering, nonvs,
241 sci_force_enable, nobl }
242 See Documentation/power/video.rst for information on
244 s3_beep is for debugging; it makes the PC's speaker beep
245 as soon as the kernel's real-mode entry point is called.
246 s4_hwsig causes the kernel to check the ACPI hardware
247 signature during resume from hibernation, and gracefully
248 refuse to resume if it has changed. This complies with
249 the ACPI specification but not with reality, since
250 Windows does not do this and many laptops do change it
251 on docking. So the default behaviour is to allow resume
252 and simply warn when the signature changes, unless the
253 s4_hwsig option is enabled.
254 s4_nohwsig prevents ACPI hardware signature from being
255 used (or even warned about) during resume.
256 old_ordering causes the ACPI 1.0 ordering of the _PTS
257 control method, with respect to putting devices into
258 low power states, to be enforced (the ACPI 2.0 ordering
259 of _PTS is used by default).
260 nonvs prevents the kernel from saving/restoring the
261 ACPI NVS memory during suspend/hibernation and resume.
262 sci_force_enable causes the kernel to set SCI_EN directly
263 on resume from S1/S3 (which is against the ACPI spec,
264 but some broken systems don't work without it).
265 nobl causes the internal blacklist of systems known to
266 behave incorrectly in some ways with respect to system
267 suspend and resume to be ignored (use wisely).
269 acpi_use_timer_override [HW,ACPI,EARLY]
270 Use timer override. For some broken Nvidia NF5 boards
271 that require a timer override, but don't have HPET
273 add_efi_memmap [EFI,X86,EARLY] Include EFI memory map in
274 kernel's map of available physical RAM.
277 { off | try_unsupported }
278 off: disable AGP support
279 try_unsupported: try to drive unsupported chipsets
280 (may crash computer or cause data corruption)
283 See Documentation/sound/alsa-configuration.rst
286 Allow the default userspace alignment fault handler
287 behaviour to be specified. Bit 0 enables warnings,
288 bit 1 enables fixups, and bit 2 sends a segfault.
290 align_va_addr= [X86-64]
291 Align virtual addresses by clearing slice [14:12] when
292 allocating a VMA at process creation time. This option
293 gives you up to 3% performance improvement on AMD F15h
294 machines (where it is enabled by default) for a
295 CPU-intensive style benchmark, and it can vary highly in
296 a microbenchmark depending on workload and compiler.
298 32: only for 32-bit processes
299 64: only for 64-bit processes
300 on: enable for both 32- and 64-bit processes
301 off: disable for both 32- and 64-bit processes
303 alloc_snapshot [FTRACE]
304 Allocate the ftrace snapshot buffer on boot up when the
305 main buffer is allocated. This is handy if debugging
306 and you need to use tracing_snapshot() on boot up, and
307 do not want to use tracing_snapshot_alloc() as it needs
308 to be done where GFP_KERNEL allocations are allowed.
310 allow_mismatched_32bit_el0 [ARM64,EARLY]
311 Allow execve() of 32-bit applications and setting of the
312 PER_LINUX32 personality on systems where only a strict
313 subset of the CPUs support 32-bit EL0. When this
314 parameter is present, the set of CPUs supporting 32-bit
315 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0
316 and hot-unplug operations may be restricted.
318 See Documentation/arch/arm64/asymmetric-32bit.rst for more
321 amd_iommu= [HW,X86-64]
322 Pass parameters to the AMD IOMMU driver in the system.
324 fullflush - Deprecated, equivalent to iommu.strict=1
325 off - do not initialize any AMD IOMMU found in
327 force_isolation - Force device isolation for all
328 devices. The IOMMU driver is not
329 allowed anymore to lift isolation
330 requirements as needed. This option
331 does not override iommu=pt
332 force_enable - Force enable the IOMMU on platforms known
333 to be buggy with IOMMU enabled. Use this
335 pgtbl_v1 - Use v1 page table for DMA-API (Default).
336 pgtbl_v2 - Use v2 page table for DMA-API.
337 irtcachedis - Disable Interrupt Remapping Table (IRT) caching.
339 amd_iommu_dump= [HW,X86-64]
340 Enable AMD IOMMU driver option to dump the ACPI table
341 for AMD IOMMU. With this option enabled, AMD IOMMU
342 driver will print ACPI tables for AMD IOMMU during
343 IOMMU initialization.
345 amd_iommu_intr= [HW,X86-64]
346 Specifies one of the following AMD IOMMU interrupt
348 legacy - Use legacy interrupt remapping mode.
349 vapic - Use virtual APIC mode, which allows IOMMU
350 to inject interrupts directly into guest.
351 This mode requires kvm-amd.avic=1.
352 (Default when IOMMU HW support is present.)
354 amd_pstate= [X86,EARLY]
356 Do not enable amd_pstate as the default
357 scaling driver for the supported processors
359 Use amd_pstate with passive mode as a scaling driver.
360 In this mode autonomous selection is disabled.
361 Driver requests a desired performance level and platform
362 tries to match the same performance level if it is
363 satisfied by guaranteed performance level.
365 Use amd_pstate_epp driver instance as the scaling driver,
366 driver provides a hint to the hardware if software wants
367 to bias toward performance (0x0) or energy efficiency (0xff)
368 to the CPPC firmware. then CPPC power algorithm will
369 calculate the runtime workload and adjust the realtime cores
372 Activate guided autonomous mode. Driver requests minimum and
373 maximum performance level and the platform autonomously
374 selects a performance level in this range and appropriate
375 to the current workload.
377 amijoy.map= [HW,JOY] Amiga joystick support
378 Map of devices attached to JOY0DAT and JOY1DAT
380 See also Documentation/input/joydev/joystick.rst
382 analog.map= [HW,JOY] Analog joystick and gamepad support
383 Specifies type or capabilities of an analog joystick
384 connected to one of 16 gameports
385 Format: <type1>,<type2>,..<type16>
388 Power management functions (SPARCstation-4/5 + deriv.)
390 Disable APC CPU standby support. SPARCstation-Fox does
391 not play well with APC CPU idle - disable it if you have
392 APC and your system crashes randomly.
394 apic= [APIC,X86,EARLY] Advanced Programmable Interrupt Controller
395 Change the output verbosity while booting
396 Format: { quiet (default) | verbose | debug }
397 Change the amount of debugging information output
398 when initialising the APIC and IO-APIC components.
399 For X86-32, this can also be used to specify an APIC
401 Format: apic=driver_name
402 Examples: apic=bigsmp
404 apic_extnmi= [APIC,X86,EARLY] External NMI delivery setting
405 Format: { bsp (default) | all | none }
406 bsp: External NMI is delivered only to CPU 0
407 all: External NMIs are broadcast to all CPUs as a
409 none: External NMI is masked for all CPUs. This is
410 useful so that a dump capture kernel won't be
414 See Documentation/networking/ipv6.rst.
416 apm= [APM] Advanced Power Management
417 See header of arch/x86/kernel/apm_32.c.
419 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
420 Format: { "0" | "1" }
421 See security/apparmor/Kconfig help text
424 Default value is set via kernel config option.
426 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
427 Format: <io>,<irq>,<nodeID>
429 arm64.nobti [ARM64] Unconditionally disable Branch Target
430 Identification support
432 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory
433 Set instructions support
435 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
438 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
441 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix
444 arm64.nosve [ARM64] Unconditionally disable Scalable Vector
449 atarimouse= [HW,MOUSE] Atari Mouse
451 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
452 EzKey and similar keyboards
454 atkbd.reset= [HW] Reset keyboard during initialization
456 atkbd.set= [HW] Select keyboard code set
457 Format: <int> (2 = AT (default), 3 = PS/2)
459 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
462 atkbd.softraw= [HW] Choose between synthetic and real raw mode
463 Format: <bool> (0 = real, 1 = synthetic (default))
465 atkbd.softrepeat= [HW]
466 Use software keyboard repeat
468 audit= [KNL] Enable the audit sub-system
469 Format: { "0" | "1" | "off" | "on" }
470 0 | off - kernel audit is disabled and can not be
471 enabled until the next reboot
472 unset - kernel audit is initialized but disabled and
473 will be fully enabled by the userspace auditd.
474 1 | on - kernel audit is initialized and partially
475 enabled, storing at most audit_backlog_limit
476 messages in RAM until it is fully enabled by the
480 audit_backlog_limit= [KNL] Set the audit queue size limit.
481 Format: <int> (must be >=0)
484 bau= [X86_UV] Enable the BAU on SGI UV. The default
485 behavior is to disable the BAU (i.e. bau=0).
486 Format: { "0" | "1" }
489 unset - Disable the BAU.
491 baycom_epp= [HW,AX25]
494 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
496 See header of drivers/net/hamradio/baycom_par.c.
498 baycom_ser_fdx= [HW,AX25]
499 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
500 Format: <io>,<irq>,<mode>[,<baud>]
501 See header of drivers/net/hamradio/baycom_ser_fdx.c.
503 baycom_ser_hdx= [HW,AX25]
504 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
505 Format: <io>,<irq>,<mode>
506 See header of drivers/net/hamradio/baycom_ser_hdx.c.
509 Disable BERT OS support on buggy BIOSes.
511 bgrt_disable [ACPI,X86,EARLY]
512 Disable BGRT to avoid flickering OEM logo.
514 blkdevparts= Manual partition parsing of block device(s) for
515 embedded devices based on command line input.
516 See Documentation/block/cmdline-partition.rst
518 boot_delay= [KNL,EARLY]
519 Milliseconds to delay each printk during boot.
520 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled,
521 and you may also have to specify "lpj=". Boot_delay
522 values larger than 10 seconds (10000) are assumed
523 erroneous and ignored.
526 bootconfig [KNL,EARLY]
527 Extended command line options can be added to an initrd
528 and this will cause the kernel to look for it.
530 See Documentation/admin-guide/bootconfig.rst
532 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
533 bttv.radio= Most important insmod options are available as
535 bttv.pll= See Documentation/admin-guide/media/bttv.rst
538 bulk_remove=off [PPC] This parameter disables the use of the pSeries
539 firmware feature for flushing multiple hpte entries
542 c101= [NET] Moxa C101 synchronous serial card
544 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
545 Sometimes CPU hardware bugs make them report the cache
546 size incorrectly. The kernel will attempt work arounds
547 to fix known problems, but for some CPUs it is not
548 possible to determine what the correct size should be.
549 This option provides an override for these situations.
552 [NET] Specifies amount of time (in seconds) that
553 the kernel should wait for a network carrier. By default
554 it waits 120 seconds.
556 ca_keys= [KEYS] This parameter identifies a specific key(s) on
557 the system trusted keyring to be used for certificate
559 format: { id:<keyid> | builtin }
561 cca= [MIPS,EARLY] Override the kernel pages' cache coherency
562 algorithm. Accepted values range from 0 to 7
563 inclusive. See arch/mips/include/asm/pgtable-bits.h
564 for platform specific values (SB1, Loongson3 and
567 ccw_timeout_log [S390]
568 See Documentation/arch/s390/common_io.rst for details.
570 cgroup_disable= [KNL] Disable a particular controller or optional feature
571 Format: {name of the controller(s) or feature(s) to disable}
572 The effects of cgroup_disable=foo are:
573 - foo isn't auto-mounted if you mount all cgroups in
575 - foo isn't visible as an individually mountable
577 - if foo is an optional feature then the feature is
578 disabled and corresponding cgroup files are not
580 {Currently only "memory" controller deal with this and
581 cut the overhead, others just disable the usage. So
582 only cgroup_disable=memory is actually worthy}
583 Specifying "pressure" disables per-cgroup pressure
584 stall information accounting feature
586 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
587 Format: { { controller | "all" | "named" }
588 [,{ controller | "all" | "named" }...] }
589 Like cgroup_disable, but only applies to cgroup v1;
590 the blacklisted controllers remain available in cgroup2.
591 "all" blacklists all controllers and "named" disables
592 named mounts. Specifying both "all" and "named" disables
595 cgroup_favordynmods= [KNL] Enable or Disable favordynmods.
596 Format: { "true" | "false" }
597 Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS.
599 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
601 nosocket -- Disable socket memory accounting.
602 nokmem -- Disable kernel memory accounting.
603 nobpf -- Disable BPF memory accounting.
605 checkreqprot= [SELINUX] Set initial checkreqprot flag value.
606 Format: { "0" | "1" }
607 See security/selinux/Kconfig help text.
608 0 -- check protection applied by kernel (includes
609 any implied execute protection).
610 1 -- check protection requested by application.
611 Default value is set via a kernel config option.
612 Value can be changed at runtime via
613 /sys/fs/selinux/checkreqprot.
614 Setting checkreqprot to 1 is deprecated.
617 See Documentation/arch/s390/common_io.rst for details.
619 clearcpuid=X[,X...] [X86]
620 Disable CPUID feature X for the kernel. See
621 arch/x86/include/asm/cpufeatures.h for the valid bit
622 numbers X. Note the Linux-specific bits are not necessarily
623 stable over kernel options, but the vendor-specific
625 X can also be a string as appearing in the flags: line
626 in /proc/cpuinfo which does not have the above
627 instability issue. However, not all features have names
629 Note that using this option will taint your kernel.
630 Also note that user programs calling CPUID directly
631 or using the feature without checking anything
632 will still see it. This just prevents it from
633 being used by the kernel or shown in /proc/cpuinfo.
634 Also note the kernel might malfunction if you disable
639 Prevents the clock framework from automatically gating
640 clocks that have not been explicitly enabled by a Linux
641 device driver but are enabled in hardware at reset or
642 by the bootloader/firmware. Note that this does not
643 force such clocks to be always-on nor does it reserve
644 those clocks in any way. This parameter is useful for
645 debug and development, but should not be needed on a
646 platform with proper driver support. For more
647 information, see Documentation/driver-api/clk.rst.
649 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
651 Forces specified clocksource (if available) to be used
652 when calculating gettimeofday(). If specified
653 clocksource is not available, it defaults to PIT.
654 Format: { pit | tsc | cyclone | pmtmr }
656 clocksource= Override the default clocksource
658 Override the default clocksource and use the clocksource
659 with the name specified.
660 Some clocksource names to choose from, depending on
662 [all] jiffies (this is the base, fallback clocksource)
664 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
665 pxa_timer,timer3,32k_counter,timer0_1
666 [X86-32] pit,hpet,tsc;
667 scx200_hrt on Geode; cyclone on IBM x440
675 clocksource.arm_arch_timer.evtstrm=
678 Enable/disable the eventstream feature of the ARM
679 architected timer so that code using WFE-based polling
680 loops can be debugged more effectively on production
683 clocksource.max_cswd_read_retries= [KNL]
684 Number of clocksource_watchdog() retries due to
685 external delays before the clock will be marked
686 unstable. Defaults to two retries, that is,
687 three attempts to read the clock under test.
689 clocksource.verify_n_cpus= [KNL]
690 Limit the number of CPUs checked for clocksources
691 marked with CLOCK_SOURCE_VERIFY_PERCPU that
692 are marked unstable due to excessive skew.
693 A negative value says to check all CPUs, while
694 zero says not to check any. Values larger than
695 nr_cpu_ids are silently truncated to nr_cpu_ids.
696 The actual CPUs are chosen randomly, with
697 no replacement if the same CPU is chosen twice.
699 clocksource-wdtest.holdoff= [KNL]
700 Set the time in seconds that the clocksource
701 watchdog test waits before commencing its tests.
702 Defaults to zero when built as a module and to
703 10 seconds when built into the kernel.
705 cma=nn[MG]@[start[MG][-end[MG]]]
707 Sets the size of kernel global memory area for
708 contiguous memory allocations and optionally the
709 placement constraint by the physical address range of
710 memory allocations. A value of 0 disables CMA
711 altogether. For more information, see
712 kernel/dma/contiguous.c
716 Sets the size of kernel per-numa memory area for
717 contiguous memory allocations. A value of 0 disables
718 per-numa CMA altogether. And If this option is not
719 specified, the default value is 0.
720 With per-numa CMA enabled, DMA users on node nid will
721 first try to allocate buffer from the pernuma area
722 which is located in node nid, if the allocation fails,
723 they will fallback to the global default memory area.
725 numa_cma=<node>:nn[MG][,<node>:nn[MG]]
727 Sets the size of kernel numa memory area for
728 contiguous memory allocations. It will reserve CMA
729 area for the specified node.
731 With numa CMA enabled, DMA users on node nid will
732 first try to allocate buffer from the numa area
733 which is located in node nid, if the allocation fails,
734 they will fallback to the global default memory area.
736 cmo_free_hint= [PPC] Format: { yes | no }
737 Specify whether pages are marked as being inactive
738 when they are freed. This is used in CMO environments
739 to determine OS memory pressure for page stealing by
743 coherent_pool=nn[KMG] [ARM,KNL,EARLY]
744 Sets the size of memory pool for coherent, atomic dma
745 allocations, by default set to 256K.
747 com20020= [HW,NET] ARCnet - COM20020 chipset
749 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
751 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
755 ARCnet - COM90xx chipset (memory-mapped buffers)
756 Format: <io>[,<irq>[,<memstart>]]
758 condev= [HW,S390] console device
761 con3215_drop= [S390,EARLY] 3215 console drop mode.
763 When set to true, drop data on the 3215 console when
764 the console buffer is full. In this case the
765 operator using a 3270 terminal emulator (for example
766 x3270) does not have to enter the clear key for the
767 console output to advance and the kernel to continue.
768 This leads to a much faster boot time when a 3270
769 terminal emulator is active. If no 3270 terminal
770 emulator is used, this parameter has no effect.
772 console= [KNL] Output console device and options.
774 tty<n> Use the virtual console device <n>.
778 Use the specified serial port. The options are of
779 the form "bbbbpnf", where "bbbb" is the baud rate,
780 "p" is parity ("n", "o", or "e"), "n" is number of
781 bits, and "f" is flow control ("r" for RTS or
782 omit it). Default is "9600n8".
784 See Documentation/admin-guide/serial-console.rst for more
786 Documentation/networking/netconsole.rst for an
789 uart[8250],io,<addr>[,options]
790 uart[8250],mmio,<addr>[,options]
791 uart[8250],mmio16,<addr>[,options]
792 uart[8250],mmio32,<addr>[,options]
793 uart[8250],0x<addr>[,options]
794 Start an early, polled-mode console on the 8250/16550
795 UART at the specified I/O port or MMIO address,
796 switching to the matching ttyS device later.
797 MMIO inter-register address stride is either 8-bit
798 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
799 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
800 to be equivalent to 'mmio'. 'options' are specified in
801 the same format described for ttyS above; if unspecified,
802 the h/w is not re-initialized.
804 hvc<n> Use the hypervisor console device <n>. This is for
805 both Xen and PowerPC hypervisors.
808 Use to disable console output, i.e., to have kernel
809 console messages discarded.
810 This must be the only console= parameter used on the
813 If the device connected to the port is not a TTY but a braille
814 device, prepend "brl," before the device type, for instance
816 For now, only VisioBraille is supported.
819 [KNL] Change console messages format
821 By default we print messages on consoles in
822 "[time stamp] text\n" format (time stamp may not be
823 printed, depending on CONFIG_PRINTK_TIME or
824 `printk_time' param).
826 Switch to syslog format: "<%u>[time stamp] text\n"
827 IOW, each message will have a facility and loglevel
828 prefix. The format is similar to one used by syslog()
829 syscall, or to executing "dmesg -S --raw" or to reading
832 consoleblank= [KNL] The console blank (screen saver) timeout in
833 seconds. A value of 0 disables the blank timer.
837 [KNL] Change the default value for
838 /proc/<pid>/coredump_filter.
839 See also Documentation/filesystems/proc.rst.
841 coresight_cpu_debug.enable
844 Enable/disable the CPU sampling based debugging.
845 0: default value, disable debugging
846 1: enable debugging at boot time
848 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
850 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
852 cpuidle.off=1 [CPU_IDLE]
853 disable the cpuidle sub-system
856 [CPU_IDLE] Name of the cpuidle governor to use.
858 cpufreq.off=1 [CPU_FREQ]
859 disable the cpufreq sub-system
861 cpufreq.default_governor=
862 [CPU_FREQ] Name of the default cpufreq governor or
863 policy to use. This governor must be registered in the
864 kernel before the cpufreq driver probes.
867 [X86,EARLY] Delay for N microsec between assert and de-assert
868 of APIC INIT to start processors. This delay occurs
869 on every CPU online, such as boot, and resume from suspend.
873 [SMP] Enable/disable parallel bringup of secondary CPUs
875 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise
876 the parameter has no effect.
878 crash_kexec_post_notifiers
879 Run kdump after running panic-notifiers and dumping
880 kmsg. This only for the users who doubt kdump always
881 succeeds in any situation.
882 Note that this also increases risks of kdump failure,
883 because some panic notifiers can make the crashed
884 kernel more unstable.
886 crashkernel=size[KMG][@offset[KMG]]
887 [KNL,EARLY] Using kexec, Linux can switch to a 'crash kernel'
888 upon panic. This parameter reserves the physical
889 memory region [offset, offset + size] for that kernel
890 image. If '@offset' is omitted, then a suitable offset
891 is selected automatically.
892 [KNL, X86-64, ARM64, RISCV, LoongArch] Select a region
893 under 4G first, and fall back to reserve region above
894 4G when '@offset' hasn't been specified.
895 See Documentation/admin-guide/kdump/kdump.rst for further details.
897 crashkernel=range1:size1[,range2:size2,...][@offset]
898 [KNL] Same as above, but depends on the memory
899 in the running system. The syntax of range is
900 start-[end] where start and end are both
901 a memory unit (amount[KMG]). See also
902 Documentation/admin-guide/kdump/kdump.rst for an example.
904 crashkernel=size[KMG],high
905 [KNL, X86-64, ARM64, RISCV, LoongArch] range could be
907 Allow kernel to allocate physical memory region from top,
908 so could be above 4G if system have more than 4G ram
909 installed. Otherwise memory region will be allocated
910 below 4G, if available.
911 It will be ignored if crashkernel=X is specified.
912 crashkernel=size[KMG],low
913 [KNL, X86-64, ARM64, RISCV, LoongArch] range under 4G.
914 When crashkernel=X,high is passed, kernel could allocate
915 physical memory region above 4G, that cause second kernel
916 crash on system that require some amount of low memory,
917 e.g. swiotlb requires at least 64M+32K low memory, also
918 enough extra low memory is needed to make sure DMA buffers
919 for 32-bit devices won't run out. Kernel would try to allocate
920 default size of memory below 4G automatically. The default
921 size is platform dependent.
922 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB)
925 --> loongarch: 128MiB
926 This one lets the user specify own low range under 4G
927 for second kernel instead.
928 0: to disable low allocation.
929 It will be ignored when crashkernel=X,high is not used
930 or memory reserved is below 4G.
933 [KNL] Disable crypto self-tests
938 cs89x0_media= [HW,NET]
939 Format: { rj45 | aui | bnc }
941 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU
942 function call handling. When switched on,
943 additional debug data is printed to the console
944 in case a hanging CPU is detected, and that
945 CPU is pinged again in order to try to resolve
946 the hang situation. The default value of this
947 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT
951 See header of drivers/s390/block/dasd_devmap.c.
953 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
954 (one device per port)
955 Format: <port#>,<type>
956 See also Documentation/input/devices/joystick-parport.rst
958 debug [KNL,EARLY] Enable kernel debugging (events log level).
961 [KNL,EARLY] Enable printing [hashed] pointers early in the
962 boot sequence. If enabled, we use a weak hash instead
963 of siphash to hash pointers. Use this option if you are
964 seeing instances of '(___ptrval___)') and need to see a
965 value (hashed pointer) instead. Cryptographically
966 insecure, please do not use on production kernels.
969 [KNL] verbose locking self-tests
971 Print debugging info while doing the locking API
973 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
974 (no extra messages), setting it to -1 (all bits set)
975 will print _a_lot_ more information - normally only
976 useful to lockdep developers.
978 debug_objects [KNL,EARLY] Enable object debugging
980 debug_guardpage_minorder=
981 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this
982 parameter allows control of the order of pages that will
983 be intentionally kept free (and hence protected) by the
984 buddy allocator. Bigger value increase the probability
985 of catching random memory corruption, but reduce the
986 amount of memory for normal system use. The maximum
987 possible value is MAX_PAGE_ORDER/2. Setting this
988 parameter to 1 or 2 should be enough to identify most
989 random memory corruption problems caused by bugs in
990 kernel or driver code when a CPU writes to (or reads
991 from) a random memory location. Note that there exists
992 a class of memory corruptions problems caused by buggy
993 H/W or F/W or by drivers badly programming DMA
994 (basically when memory is written at bus level and the
995 CPU MMU is bypassed) which are not detectable by
996 CONFIG_DEBUG_PAGEALLOC, hence this option will not
997 help tracking down these problems.
1000 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
1001 enables the feature at boot time. By default, it is
1002 disabled and the system will work mostly the same as a
1003 kernel built without CONFIG_DEBUG_PAGEALLOC.
1004 Note: to get most of debug_pagealloc error reports, it's
1005 useful to also enable the page_owner functionality.
1006 on: enable the feature
1008 debugfs= [KNL,EARLY] This parameter enables what is exposed to
1009 userspace and debugfs internal clients.
1010 Format: { on, no-mount, off }
1011 on: All functions are enabled.
1013 Filesystem is not registered but kernel clients can
1014 access APIs and a crashkernel can be used to read
1015 its content. There is nothing to mount.
1016 off: Filesystem is not registered and clients
1017 get a -EPERM as result when trying to register files
1018 or directories within debugfs.
1019 This is equivalent of the runtime functionality if
1020 debugfs was not enabled in the kernel at all.
1021 Default value is set in build-time with a kernel configuration.
1023 debugpat [X86] Enable PAT debugging
1026 [HW] The size of the default HugeTLB page. This is
1027 the size represented by the legacy /proc/ hugepages
1028 APIs. In addition, this is the default hugetlb size
1029 used for shmget(), mmap() and mounting hugetlbfs
1030 filesystems. If not specified, defaults to the
1031 architecture's default huge page size. Huge page
1032 sizes are architecture dependent. See also
1033 Documentation/admin-guide/mm/hugetlbpage.rst.
1036 deferred_probe_timeout=
1037 [KNL] Debugging option to set a timeout in seconds for
1038 deferred probe to give up waiting on dependencies to
1039 probe. Only specific dependencies (subsystems or
1040 drivers) that have opted in will be ignored. A timeout
1041 of 0 will timeout at the end of initcalls. If the time
1042 out hasn't expired, it'll be restarted by each
1043 successful driver registration. This option will also
1044 dump out devices still on the deferred probe list after
1047 delayacct [KNL] Enable per-task delay accounting
1049 dell_smm_hwmon.ignore_dmi=
1050 [HW] Continue probing hardware even if DMI data
1051 indicates that the driver is running on unsupported
1054 dell_smm_hwmon.force=
1055 [HW] Activate driver even if SMM BIOS signature does
1056 not match list of supported models and enable otherwise
1057 blacklisted features.
1059 dell_smm_hwmon.power_status=
1060 [HW] Report power status in /proc/i8k
1061 (disabled by default).
1063 dell_smm_hwmon.restricted=
1064 [HW] Allow controlling fans only if SYS_ADMIN
1067 dell_smm_hwmon.fan_mult=
1068 [HW] Factor to multiply fan speed with.
1070 dell_smm_hwmon.fan_max=
1071 [HW] Maximum configurable fan speed.
1074 Format: { on | off | def_only | inf_only | always }
1075 on: s390 zlib hardware support for compression on
1076 level 1 and decompression (default)
1077 off: No s390 zlib hardware support
1078 def_only: s390 zlib hardware support for deflate
1079 only (compression on level 1)
1080 inf_only: s390 zlib hardware support for inflate
1081 only (decompression)
1082 always: Same as 'on' but ignores the selected compression
1083 level always using hardware support (used for debugging)
1085 dhash_entries= [KNL]
1086 Set number of hash buckets for dentry cache.
1088 disable_1tb_segments [PPC,EARLY]
1089 Disables the use of 1TB hash page table segments. This
1090 causes the kernel to fall back to 256MB segments which
1091 can be useful when debugging issues that require an SLB
1095 See Documentation/networking/ipv6.rst.
1097 disable_radix [PPC,EARLY]
1098 Disable RADIX MMU mode on POWER9
1101 Disable TLBIE instruction. Currently does not work
1102 with KVM, with HASH MMU, or with coherent accelerators.
1104 disable_cpu_apicid= [X86,APIC,SMP]
1106 The number of initial APIC ID for the
1107 corresponding CPU to be disabled at boot,
1108 mostly used for the kdump 2nd kernel to
1109 disable BSP to wake up multiple CPUs without
1110 causing system reset or hang due to sending
1111 INIT from AP to BSP.
1113 disable_ddw [PPC/PSERIES,EARLY]
1114 Disable Dynamic DMA Window support. Use this
1115 to workaround buggy firmware.
1117 disable_ipv6= [IPV6]
1118 See Documentation/networking/ipv6.rst.
1120 disable_mtrr_cleanup [X86,EARLY]
1121 The kernel tries to adjust MTRR layout from continuous
1122 to discrete, to make X server driver able to add WB
1123 entry later. This parameter disables that.
1125 disable_mtrr_trim [X86, Intel and AMD only,EARLY]
1126 By default the kernel will trim any uncacheable
1127 memory out of your available memory pool based on
1128 MTRR settings. This parameter disables that behavior,
1129 possibly causing your machine to run very slowly.
1131 disable_timer_pin_1 [X86,EARLY]
1132 Disable PIN 1 of APIC timer
1133 Can be useful to work around chipset bugs.
1135 dis_ucode_ldr [X86] Disable the microcode loader.
1137 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1138 this option disables the debugging code at boot.
1140 dma_debug_entries=<number>
1141 This option allows to tune the number of preallocated
1142 entries for DMA-API debugging code. One entry is
1143 required per DMA-API allocation. Use this if the
1144 DMA-API debugging code disables itself because the
1145 architectural default is too low.
1147 dma_debug_driver=<driver_name>
1148 With this option the DMA-API debugging driver
1149 filter feature can be enabled at boot time. Just
1150 pass the driver to filter for as the parameter.
1151 The filter can be disabled or changed to another
1152 driver later using sysfs.
1154 driver_async_probe= [KNL]
1155 List of driver names to be probed asynchronously. *
1156 matches with all driver names. If * is specified, the
1157 rest of the listed driver names are those that will NOT
1159 Format: <driver_name1>,<driver_name2>...
1161 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1162 Broken monitors, graphic adapters, KVMs and EDIDless
1163 panels may send no or incorrect EDID data sets.
1164 This parameter allows to specify an EDID data sets
1165 in the /lib/firmware directory that are used instead.
1166 Generic built-in EDID data sets are used, if one of
1167 edid/1024x768.bin, edid/1280x1024.bin,
1168 edid/1680x1050.bin, or edid/1920x1080.bin is given
1169 and no file with the same name exists. Details and
1170 instructions how to build your own EDID data are
1171 available in Documentation/admin-guide/edid.rst. An EDID
1172 data set will only be used for a particular connector,
1173 if its name and a colon are prepended to the EDID
1174 name. Each connector may use a unique EDID data
1175 set by separating the files with a comma. An EDID
1176 data set with no connector name will be used for
1177 any connectors not explicitly specified.
1181 dt_cpu_ftrs= [PPC,EARLY]
1182 Format: {"off" | "known"}
1183 Control how the dt_cpu_ftrs device-tree binding is
1184 used for CPU feature discovery and setup (if it
1186 off: Do not use it, fall back to legacy cpu table.
1187 known: Do not pass through unknown features to guests
1188 or userspace, only those that the kernel is aware of.
1190 dump_apple_properties [X86]
1191 Dump name and content of EFI device properties on
1192 x86 Macs. Useful for driver authors to determine
1193 what data is available or for reverse-engineering.
1195 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1196 <module>.dyndbg[="val"]
1197 Enable debug messages at boot time. See
1198 Documentation/admin-guide/dynamic-debug-howto.rst
1201 early_ioremap_debug [KNL,EARLY]
1202 Enable debug messages in early_ioremap support. This
1203 is useful for tracking down temporary early mappings
1204 which are not unmapped.
1206 earlycon= [KNL,EARLY] Output early console device and options.
1208 When used with no options, the early console is
1209 determined by stdout-path property in device tree's
1210 chosen node or the ACPI SPCR table if supported by
1213 cdns,<addr>[,options]
1214 Start an early, polled-mode console on a Cadence
1215 (xuartps) serial port at the specified address. Only
1216 supported option is baud rate. If baud rate is not
1217 specified, the serial port must already be setup and
1220 uart[8250],io,<addr>[,options[,uartclk]]
1221 uart[8250],mmio,<addr>[,options[,uartclk]]
1222 uart[8250],mmio32,<addr>[,options[,uartclk]]
1223 uart[8250],mmio32be,<addr>[,options[,uartclk]]
1224 uart[8250],0x<addr>[,options]
1225 Start an early, polled-mode console on the 8250/16550
1226 UART at the specified I/O port or MMIO address.
1227 MMIO inter-register address stride is either 8-bit
1228 (mmio) or 32-bit (mmio32 or mmio32be).
1229 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1230 to be equivalent to 'mmio'. 'options' are specified
1231 in the same format described for "console=ttyS<n>"; if
1232 unspecified, the h/w is not initialized. 'uartclk' is
1233 the uart clock frequency; if unspecified, it is set
1234 to 'BASE_BAUD' * 16.
1238 Start an early, polled-mode console on a pl011 serial
1239 port at the specified address. The pl011 serial port
1240 must already be setup and configured. Options are not
1241 yet supported. If 'mmio32' is specified, then only
1242 the driver will use only 32-bit accessors to read/write
1243 the device registers.
1246 Start an early console on a litex serial port at the
1247 specified address. The serial port must already be
1248 setup and configured. Options are not yet supported.
1251 Start an early, polled-mode console on a meson serial
1252 port at the specified address. The serial port must
1253 already be setup and configured. Options are not yet
1257 Start an early, polled-mode console on an msm serial
1258 port at the specified address. The serial port
1259 must already be setup and configured. Options are not
1262 msm_serial_dm,<addr>
1263 Start an early, polled-mode console on an msm serial
1264 dm port at the specified address. The serial port
1265 must already be setup and configured. Options are not
1269 Start an early, polled-mode console on a serial port
1270 of an Actions Semi SoC, such as S500 or S900, at the
1271 specified address. The serial port must already be
1272 setup and configured. Options are not yet supported.
1275 Start an early, polled-mode console on a serial port
1276 of an RDA Micro SoC, such as RDA8810PL, at the
1277 specified address. The serial port must already be
1278 setup and configured. Options are not yet supported.
1281 Use RISC-V SBI (Supervisor Binary Interface) for early
1284 smh Use ARM semihosting calls for early console.
1292 Use early console provided by serial driver available
1293 on Samsung SoCs, requires selecting proper type and
1294 a correct base address of the selected UART port. The
1295 serial port must already be setup and configured.
1296 Options are not yet supported.
1299 Start an early, polled-mode console on a lantiq serial
1300 (lqasc) port at the specified address. The serial port
1301 must already be setup and configured. Options are not
1306 Use early console provided by Freescale LP UART driver
1307 found on Freescale Vybrid and QorIQ LS1021A processors.
1308 A valid base address must be provided, and the serial
1309 port must already be setup and configured.
1313 Start an early, polled-mode, output-only console on the
1314 Freescale i.MX UART at the specified address. The UART
1315 must already be setup and configured.
1318 Start an early, polled-mode console on the
1319 Armada 3700 serial port at the specified
1320 address. The serial port must already be setup
1321 and configured. Options are not yet supported.
1324 Start an early, polled-mode console on a Qualcomm
1325 Generic Interface (GENI) based serial port at the
1326 specified address. The serial port must already be
1327 setup and configured. Options are not yet supported.
1330 Start an early, unaccelerated console on the EFI
1331 memory mapped framebuffer (if available). On cache
1332 coherent non-x86 systems that use system memory for
1333 the framebuffer, pass the 'ram' option so that it is
1334 mapped with the correct attributes.
1337 Use early console provided by Freescale LINFlexD UART
1338 serial driver for NXP S32V234 SoCs. A valid base
1339 address must be provided, and the serial port must
1340 already be setup and configured.
1342 earlyprintk= [X86,SH,ARM,M68k,S390,UM,EARLY]
1346 earlyprintk=serial[,ttySn[,baudrate]]
1347 earlyprintk=serial[,0x...[,baudrate]]
1348 earlyprintk=ttySn[,baudrate]
1349 earlyprintk=dbgp[debugController#]
1350 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1351 earlyprintk=xdbc[xhciController#]
1354 earlyprintk is useful when the kernel crashes before
1355 the normal console is initialized. It is not enabled by
1356 default because it has some cosmetic problems.
1358 Append ",keep" to not disable it when the real console
1361 Only one of vga, serial, or usb debug port can
1364 Currently only ttyS0 and ttyS1 may be specified by
1365 name. Other I/O ports may be explicitly specified
1366 on some architectures (x86 and arm at least) by
1367 replacing ttySn with an I/O port address, like this:
1368 earlyprintk=serial,0x1008,115200
1369 You can find the port for a given device in
1370 /proc/tty/driver/serial:
1371 2: uart:ST16650V2 port:00001008 irq:18 ...
1373 Interaction with the standard serial driver is not
1376 The VGA output is eventually overwritten by
1379 The xen option can only be used in Xen domains.
1381 The sclp output can only be used on s390.
1383 The bios output can only be used on SuperH.
1385 The optional "force" to "pciserial" enables use of a
1386 PCI device even when its classcode is not of the
1389 edac_report= [HW,EDAC] Control how to report EDAC event
1390 Format: {"on" | "off" | "force"}
1391 on: enable EDAC to report H/W event. May be overridden
1392 by other higher priority error reporting module.
1393 off: disable H/W event reporting through EDAC.
1394 force: enforce the use of EDAC to report H/W event.
1398 Format: {"off" | "on" | "skip[mbr]"}
1401 Format: { "debug", "disable_early_pci_dma",
1402 "nochunk", "noruntime", "nosoftreserve",
1403 "novamap", "no_disable_early_pci_dma" }
1404 debug: enable misc debug output.
1405 disable_early_pci_dma: disable the busmaster bit on all
1406 PCI bridges while in the EFI boot stub.
1407 nochunk: disable reading files in "chunks" in the EFI
1408 boot stub, as chunking can cause problems with some
1409 firmware implementations.
1410 noruntime : disable EFI runtime services support
1411 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1412 attribute may cause the kernel to reserve the
1413 memory range for a memory mapping driver to
1414 claim. Specify efi=nosoftreserve to disable this
1415 reservation and treat the memory by its base type
1416 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1417 novamap: do not call SetVirtualAddressMap().
1418 no_disable_early_pci_dma: Leave the busmaster bit set
1419 on all PCI bridges while in the EFI boot stub
1421 efi_no_storage_paranoia [EFI,X86,EARLY]
1422 Using this parameter you can use more than 50% of
1423 your efi variable storage. Use this parameter only if
1424 you are really sure that your UEFI does sane gc and
1425 fulfills the spec otherwise your board may brick.
1427 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI,X86,EARLY]
1428 Add arbitrary attribute to specific memory range by
1429 updating original EFI memory map.
1430 Region of memory which aa attribute is added to is
1433 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1434 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1435 attribute is added to range 0x100000000-0x180000000 and
1436 0x10a0000000-0x1120000000.
1438 If efi_fake_mem=8G@9G:0x40000 is specified, the
1439 EFI_MEMORY_SP(0x40000) attribute is added to
1440 range 0x240000000-0x43fffffff.
1442 Using this parameter you can do debugging of EFI memmap
1443 related features. For example, you can do debugging of
1444 Address Range Mirroring feature even if your box
1445 doesn't support it, or mark specific memory as
1448 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1449 that is to be dynamically loaded by Linux. If there are
1450 multiple variables with the same name but with different
1451 vendor GUIDs, all of them will be loaded. See
1452 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1455 eisa_irq_edge= [PARISC,HW]
1456 See header of drivers/parisc/eisa.c.
1458 ekgdboc= [X86,KGDB,EARLY] Allow early kernel console debugging
1461 This is designed to be used in conjunction with
1462 the boot argument: earlyprintk=vga
1464 This parameter works in place of the kgdboc parameter
1465 but can only be used if the backing tty is available
1466 very early in the boot process. For early debugging
1467 via a serial port see kgdboc_earlycon instead.
1470 See comment before function elanfreq_setup() in
1471 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1473 elfcorehdr=[size[KMG]@]offset[KMG] [PPC,SH,X86,S390,EARLY]
1474 Specifies physical address of start of kernel core
1475 image elf header and optionally the size. Generally
1476 kexec loader will pass this option to capture kernel.
1477 See Documentation/admin-guide/kdump/kdump.rst for details.
1479 enable_mtrr_cleanup [X86,EARLY]
1480 The kernel tries to adjust MTRR layout from continuous
1481 to discrete, to make X server driver able to add WB
1482 entry later. This parameter enables that.
1484 enable_timer_pin_1 [X86]
1485 Enable PIN 1 of APIC timer
1486 Can be useful to work around chipset bugs
1487 (in particular on some ATI chipsets).
1488 The kernel tries to set a reasonable default.
1490 enforcing= [SELINUX] Set initial enforcing status.
1492 See security/selinux/Kconfig help text.
1493 0 -- permissive (log only, no denials).
1494 1 -- enforcing (deny and log).
1496 Value can be changed at runtime via
1497 /sys/fs/selinux/enforce.
1500 Disable Error Record Serialization Table (ERST)
1503 ether= [HW,NET] Ethernet cards parameters
1504 This option is obsoleted by the "netdev=" option, which
1505 has equivalent usage. See its documentation for details.
1509 Permit 'security.evm' to be updated regardless of
1510 current integrity status.
1512 early_page_ext [KNL,EARLY] Enforces page_ext initialization to earlier
1513 stages so cover more early boot allocations.
1514 Please note that as side effect some optimizations
1515 might be disabled to achieve that (e.g. parallelized
1516 memory initialization is disabled) so the boot process
1517 might take longer, especially on systems with a lot of
1518 memory. Available with CONFIG_PAGE_EXTENSION=y.
1523 fail_make_request=[KNL]
1524 General fault injection mechanism.
1525 Format: <interval>,<probability>,<space>,<times>
1526 See also Documentation/fault-injection/.
1529 Format: { initns | none }
1530 See Documentation/admin-guide/sysctl/net.rst for
1531 fb_tunnels_only_for_init_ns
1534 See Documentation/admin-guide/blockdev/floppy.rst.
1537 Forcefully enable Physical Address Extension (PAE).
1538 Many Pentium M systems disable PAE but may have a
1539 functionally usable PAE implementation.
1540 Warning: use of this parameter will taint the kernel
1541 and may cause unknown problems.
1544 [FTRACE] will set and start the specified tracer
1545 as early as possible in order to facilitate early
1548 ftrace_boot_snapshot
1549 [FTRACE] On boot up, a snapshot will be taken of the
1550 ftrace ring buffer that can be read at:
1551 /sys/kernel/tracing/snapshot.
1552 This is useful if you need tracing information from kernel
1553 boot up that is likely to be overridden by user space
1554 start up functionality.
1556 Optionally, the snapshot can also be defined for a tracing
1557 instance that was created by the trace_instance= command
1560 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo
1562 The above will cause the "foo" tracing instance to trigger
1563 a snapshot at the end of boot up.
1565 ftrace_dump_on_oops[=orig_cpu]
1566 [FTRACE] will dump the trace buffers on oops.
1567 If no parameter is passed, ftrace will dump
1568 buffers of all CPUs, but if you pass orig_cpu, it will
1569 dump only the buffer of the CPU that triggered the
1572 ftrace_filter=[function-list]
1573 [FTRACE] Limit the functions traced by the function
1574 tracer at boot up. function-list is a comma-separated
1575 list of functions. This list can be changed at run
1576 time by the set_ftrace_filter file in the debugfs
1579 ftrace_notrace=[function-list]
1580 [FTRACE] Do not trace the functions specified in
1581 function-list. This list can be changed at run time
1582 by the set_ftrace_notrace file in the debugfs
1585 ftrace_graph_filter=[function-list]
1586 [FTRACE] Limit the top level callers functions traced
1587 by the function graph tracer at boot up.
1588 function-list is a comma-separated list of functions
1589 that can be changed at run time by the
1590 set_graph_function file in the debugfs tracing directory.
1592 ftrace_graph_notrace=[function-list]
1593 [FTRACE] Do not trace from the functions specified in
1594 function-list. This list is a comma-separated list of
1595 functions that can be changed at run time by the
1596 set_graph_notrace file in the debugfs tracing directory.
1598 ftrace_graph_max_depth=<uint>
1599 [FTRACE] Used with the function graph tracer. This is
1600 the max depth it will trace into a function. This value
1601 can be changed at run time by the max_graph_depth file
1602 in the tracefs tracing directory. default: 0 (no limit)
1604 fw_devlink= [KNL,EARLY] Create device links between consumer and supplier
1605 devices by scanning the firmware to infer the
1606 consumer/supplier relationships. This feature is
1607 especially useful when drivers are loaded as modules as
1608 it ensures proper ordering of tasks like device probing
1609 (suppliers first, then consumers), supplier boot state
1610 clean up (only after all consumers have probed),
1611 suspend/resume & runtime PM (consumers first, then
1613 Format: { off | permissive | on | rpm }
1614 off -- Don't create device links from firmware info.
1615 permissive -- Create device links from firmware info
1616 but use it only for ordering boot state clean
1617 up (sync_state() calls).
1618 on -- Create device links from firmware info and use it
1619 to enforce probe and suspend/resume ordering.
1620 rpm -- Like "on", but also use to order runtime PM.
1622 fw_devlink.strict=<bool>
1623 [KNL,EARLY] Treat all inferred dependencies as mandatory
1624 dependencies. This only applies for fw_devlink=on|rpm.
1627 fw_devlink.sync_state =
1628 [KNL,EARLY] When all devices that could probe have finished
1629 probing, this parameter controls what to do with
1630 devices that haven't yet received their sync_state()
1632 Format: { strict | timeout }
1633 strict -- Default. Continue waiting on consumers to
1635 timeout -- Give up waiting on consumers and call
1636 sync_state() on any devices that haven't yet
1637 received their sync_state() calls after
1638 deferred_probe_timeout has expired or by
1639 late_initcall() if !CONFIG_MODULES.
1642 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1643 support via parallel port (up to 5 devices per port)
1644 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1645 See also Documentation/input/devices/joystick-parport.rst
1649 gart_fix_e820= [X86-64,EARLY] disable the fix e820 for K8 GART
1653 gather_data_sampling=
1654 [X86,INTEL,EARLY] Control the Gather Data Sampling (GDS)
1657 Gather Data Sampling is a hardware vulnerability which
1658 allows unprivileged speculative access to data which was
1659 previously stored in vector registers.
1661 This issue is mitigated by default in updated microcode.
1662 The mitigation may have a performance impact but can be
1663 disabled. On systems without the microcode mitigation
1664 disabling AVX serves as a mitigation.
1666 force: Disable AVX to mitigate systems without
1667 microcode mitigation. No effect if the microcode
1668 mitigation is present. Known to cause crashes in
1669 userspace with buggy AVX enumeration.
1671 off: Disable GDS mitigation.
1673 gcov_persist= [GCOV] When non-zero (default), profiling data for
1674 kernel modules is saved and remains accessible via
1675 debugfs, even when the module is unloaded/reloaded.
1676 When zero, profiling data is discarded and associated
1677 debugfs files are removed at module unload time.
1679 goldfish [X86] Enable the goldfish android emulator platform.
1680 Don't use this when you are not running on the
1683 gpio-mockup.gpio_mockup_ranges
1684 [HW] Sets the ranges of gpiochip of for this device.
1685 Format: <start1>,<end1>,<start2>,<end2>...
1686 gpio-mockup.gpio_mockup_named_lines
1687 [HW] Let the driver know GPIO lines should be named.
1689 gpt [EFI] Forces disk with valid GPT signature but
1690 invalid Protective MBR to be treated as GPT. If the
1691 primary GPT is corrupted, it enables the backup/alternate
1692 GPT to be used instead.
1694 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1695 the "Enable 0" bit of the configuration register.
1698 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1699 the "Enable 0" bit of the configuration register.
1702 grcan.select= [HW] Select which physical interface to use.
1705 grcan.txsize= [HW] Sets the size of the tx buffer.
1706 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1708 grcan.rxsize= [HW] Sets the size of the rx buffer.
1709 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1713 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
1714 hardening is enabled for this boot. Hardened
1715 usercopy checking is used to protect the kernel
1716 from reading or writing beyond known memory
1717 allocation boundaries as a proactive defense
1718 against bounds-checking flaws in the kernel's
1719 copy_to_user()/copy_from_user() interface.
1720 on Perform hardened usercopy checks (default).
1721 off Disable hardened usercopy checks.
1723 hardlockup_all_cpu_backtrace=
1724 [KNL] Should the hard-lockup detector generate
1725 backtraces on all cpus.
1728 hashdist= [KNL,NUMA] Large hashes allocated during boot
1729 are distributed across NUMA nodes. Defaults on
1730 for 64-bit NUMA, off otherwise.
1731 Format: 0 | 1 (for off | on)
1733 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1735 hd= [EIDE] (E)IDE hard drive subsystem geometry
1736 Format: <cyl>,<head>,<sect>
1739 Disable Hardware Error Source Table (HEST) support;
1740 corresponding firmware-first mode error processing
1741 logic will be disabled.
1743 hibernate= [HIBERNATION]
1744 noresume Don't check if there's a hibernation image
1745 present during boot.
1746 nocompress Don't compress/decompress hibernation images.
1747 no Disable hibernation and resume.
1748 protect_image Turn on image protection during restoration
1749 (that will set all pages holding image data
1750 during restoration read-only).
1752 highmem=nn[KMG] [KNL,BOOT,EARLY] forces the highmem zone to have an exact
1753 size of <nn>. This works even on boxes that have no
1754 highmem otherwise. This also works to reduce highmem
1755 size on bigger boxes.
1757 highres= [KNL] Enable/disable high resolution timer mode.
1758 Valid parameters: "on", "off"
1763 hostname= [KNL,EARLY] Set the hostname (aka UTS nodename).
1765 This allows setting the system's hostname during early
1766 startup. This sets the name returned by gethostname.
1767 Using this parameter to set the hostname makes it
1768 possible to ensure the hostname is correctly set before
1769 any userspace processes run, avoiding the possibility
1770 that a process may call gethostname before the hostname
1771 has been explicitly set, resulting in the calling
1772 process getting an incorrect result. The string must
1773 not exceed the maximum allowed hostname length (usually
1774 64 characters) and will be truncated otherwise.
1776 hpet= [X86-32,HPET] option to control HPET usage
1777 Format: { enable (default) | disable | force |
1779 disable: disable HPET and use PIT instead
1780 force: allow force enabled of undocumented chips (ICH4,
1782 verbose: show contents of HPET registers during setup
1784 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1785 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1787 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1788 If this follows hugepagesz (below), it specifies
1789 the number of pages of hugepagesz to be allocated.
1790 If this is the first HugeTLB parameter on the command
1791 line, it specifies the number of pages to allocate for
1792 the default huge page size. If using node format, the
1793 number of pages to allocate per-node can be specified.
1794 See also Documentation/admin-guide/mm/hugetlbpage.rst.
1795 Format: <integer> or (node format)
1796 <node>:<integer>[,<node>:<integer>]
1799 [HW] The size of the HugeTLB pages. This is used in
1800 conjunction with hugepages (above) to allocate huge
1801 pages of a specific size at boot. The pair
1802 hugepagesz=X hugepages=Y can be specified once for
1803 each supported huge page size. Huge page sizes are
1804 architecture dependent. See also
1805 Documentation/admin-guide/mm/hugetlbpage.rst.
1808 hugetlb_cma= [HW,CMA,EARLY] The size of a CMA area used for allocation
1809 of gigantic hugepages. Or using node format, the size
1810 of a CMA area per node can be specified.
1811 Format: nn[KMGTPE] or (node format)
1812 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]]
1814 Reserve a CMA area of given size and allocate gigantic
1815 hugepages using the CMA allocator. If enabled, the
1816 boot-time allocation of gigantic hugepages is skipped.
1818 hugetlb_free_vmemmap=
1819 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
1821 Control if HugeTLB Vmemmap Optimization (HVO) is enabled.
1822 Allows heavy hugetlb users to free up some more
1823 memory (7 * PAGE_SIZE for each 2MB hugetlb page).
1824 Format: { on | off (default) }
1829 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y,
1832 Note that the vmemmap pages may be allocated from the added
1833 memory block itself when memory_hotplug.memmap_on_memory is
1834 enabled, those vmemmap pages cannot be optimized even if this
1835 feature is enabled. Other vmemmap pages not allocated from
1836 the added memory block itself do not be affected.
1839 [KNL] Should the hung task detector generate panics.
1842 A value of 1 instructs the kernel to panic when a
1843 hung task is detected. The default value is controlled
1844 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1845 option. The value selected by this boot parameter can
1846 be changed later by the kernel.hung_task_panic sysctl.
1848 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1849 terminal devices. Valid values: 0..8
1850 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1851 If specified, z/VM IUCV HVC accepts connections
1852 from listed z/VM user IDs only.
1854 hv_nopvspin [X86,HYPER_V,EARLY]
1855 Disables the paravirt spinlock optimizations
1856 which allow the hypervisor to 'idle' the guest
1859 i2c_bus= [HW] Override the default board specific I2C bus speed
1860 or register an additional I2C bus that is not
1861 registered from board initialization code.
1865 i8042.debug [HW] Toggle i8042 debug mode
1866 i8042.unmask_kbd_data
1867 [HW] Enable printing of interrupt data from the KBD port
1868 (disabled by default, and as a pre-condition
1869 requires that i8042.debug=1 be enabled)
1870 i8042.direct [HW] Put keyboard port into non-translated mode
1871 i8042.dumbkbd [HW] Pretend that controller can only read data from
1872 keyboard and cannot control its state
1873 (Don't attempt to blink the leds)
1874 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1875 i8042.nokbd [HW] Don't check/create keyboard port
1876 i8042.noloop [HW] Disable the AUX Loopback command while probing
1878 i8042.nomux [HW] Don't check presence of an active multiplexing
1880 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1882 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1883 i8042.reset [HW] Reset the controller during init, cleanup and
1884 suspend-to-ram transitions, only during s2r
1885 transitions, or never reset
1886 Format: { 1 | Y | y | 0 | N | n }
1887 1, Y, y: always reset controller
1888 0, N, n: don't ever reset controller
1889 Default: only on s2r transitions on x86; most other
1890 architectures force reset to be always executed
1891 i8042.unlock [HW] Unlock (ignore) the keylock
1892 i8042.kbdreset [HW] Reset device connected to KBD port
1894 [HW] Allow deferred probing upon i8042 probe errors
1898 i915.invert_brightness=
1899 [DRM] Invert the sense of the variable that is used to
1900 set the brightness of the panel backlight. Normally a
1901 brightness value of 0 indicates backlight switched off,
1902 and the maximum of the brightness value sets the backlight
1903 to maximum brightness. If this parameter is set to 0
1904 (default) and the machine requires it, or this parameter
1905 is set to 1, a brightness value of 0 sets the backlight
1906 to maximum brightness, and the maximum of the brightness
1907 value switches the backlight off.
1908 -1 -- never invert brightness
1909 0 -- machine default
1910 1 -- force brightness inversion
1912 ia32_emulation= [X86-64]
1914 When true, allows loading 32-bit programs and executing 32-bit
1915 syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at
1916 boot time. When false, unconditionally disables IA32 emulation.
1919 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1923 Format: idle=poll, idle=halt, idle=nomwait
1924 Poll forces a polling idle loop that can slightly
1925 improve the performance of waking up a idle CPU, but
1926 will use a lot of power and make the system run hot.
1928 idle=halt: Halt is forced to be used for CPU idle.
1929 In such case C2/C3 won't be used again.
1930 idle=nomwait: Disable mwait for CPU C-states
1934 Allow force disabling of Shared Virtual Memory (SVA)
1935 support for the idxd driver. By default it is set to
1938 idxd.tc_override= [HW]
1940 Allow override of default traffic class configuration
1941 for the device. By default it is set to false (0).
1943 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1944 Format: { strict | legacy | 2008 | relaxed }
1947 Choose which programs will be accepted for execution
1948 based on the IEEE 754 NaN encoding(s) supported by
1949 the FPU and the NaN encoding requested with the value
1950 of an ELF file header flag individually set by each
1951 binary. Hardware implementations are permitted to
1952 support either or both of the legacy and the 2008 NaN
1955 Available settings are as follows:
1956 strict accept binaries that request a NaN encoding
1957 supported by the FPU
1958 legacy only accept legacy-NaN binaries, if supported
1960 2008 only accept 2008-NaN binaries, if supported
1962 relaxed accept any binaries regardless of whether
1963 supported by the FPU
1965 The FPU emulator is always able to support both NaN
1966 encodings, so if no FPU hardware is present or it has
1967 been disabled with 'nofpu', then the settings of
1968 'legacy' and '2008' strap the emulator accordingly,
1969 'relaxed' straps the emulator for both legacy-NaN and
1970 2008-NaN, whereas 'strict' enables legacy-NaN only on
1971 legacy processors and both NaN encodings on MIPS32 or
1974 The setting for ABS.fmt/NEG.fmt instruction execution
1975 mode generally follows that for the NaN encoding,
1976 except where unsupported by hardware.
1978 ignore_loglevel [KNL,EARLY]
1979 Ignore loglevel setting - this will print /all/
1980 kernel messages to the console. Useful for debugging.
1981 We also add it as printk module parameter, so users
1982 could change it dynamically, usually by
1983 /sys/module/printk/parameters/ignore_loglevel.
1986 Ignore RLIMIT_DATA setting for data mappings,
1987 print warning at first misuse. Can be changed via
1988 /sys/module/kernel/parameters/ignore_rlimit_data.
1990 ihash_entries= [KNL]
1991 Set number of hash buckets for inode cache.
1993 ima_appraise= [IMA] appraise integrity measurements
1994 Format: { "off" | "enforce" | "fix" | "log" }
1997 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1998 The builtin appraise policy appraises all files
2001 ima_canonical_fmt [IMA]
2002 Use the canonical format for the binary runtime
2003 measurements, instead of host native format.
2006 Format: { md5 | sha1 | rmd160 | sha256 | sha384
2010 The list of supported hash algorithms is defined
2011 in crypto/hash_info.h.
2014 The builtin policies to load during IMA setup.
2015 Format: "tcb | appraise_tcb | secure_boot |
2016 fail_securely | critical_data"
2018 The "tcb" policy measures all programs exec'd, files
2019 mmap'd for exec, and all files opened with the read
2020 mode bit set by either the effective uid (euid=0) or
2023 The "appraise_tcb" policy appraises the integrity of
2024 all files owned by root.
2026 The "secure_boot" policy appraises the integrity
2027 of files (eg. kexec kernel image, kernel modules,
2028 firmware, policy, etc) based on file signatures.
2030 The "fail_securely" policy forces file signature
2031 verification failure also on privileged mounted
2032 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
2035 The "critical_data" policy measures kernel integrity
2038 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
2039 Load a policy which meets the needs of the Trusted
2040 Computing Base. This means IMA will measure all
2041 programs exec'd, files mmap'd for exec, and all files
2042 opened for read by uid=0.
2045 Select one of defined IMA measurements template formats.
2046 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" |
2051 [IMA] Define a custom template format.
2052 Format: { "field1|...|fieldN" }
2054 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
2055 Format: <min_file_size>
2056 Set the minimal file size for using asynchronous hash.
2057 If left unspecified, ahash usage is disabled.
2059 ahash performance varies for different data sizes on
2060 different crypto accelerators. This option can be used
2061 to achieve the best performance for a particular HW.
2063 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
2065 Set hashing buffer size. Default: 4k.
2067 ahash performance varies for different chunk sizes on
2068 different crypto accelerators. This option can be used
2069 to achieve best performance for particular HW.
2073 Run specified binary instead of /sbin/init as init
2076 initcall_debug [KNL] Trace initcalls as they are executed. Useful
2077 for working out where the kernel is dying during
2080 initcall_blacklist= [KNL] Do not execute a comma-separated list of
2081 initcall functions. Useful for debugging built-in
2082 modules and initcalls.
2084 initramfs_async= [KNL]
2087 This parameter controls whether the initramfs
2088 image is unpacked asynchronously, concurrently
2089 with devices being probed and
2090 initialized. This should normally just work,
2091 but as a debugging aid, one can get the
2092 historical behaviour of the initramfs
2093 unpacking being completed before device_ and
2096 initrd= [BOOT,EARLY] Specify the location of the initial ramdisk
2098 initrdmem= [KNL,EARLY] Specify a physical address and size from which to
2099 load the initrd. If an initrd is compiled in or
2100 specified in the bootparams, it takes priority over this
2102 Format: ss[KMG],nn[KMG]
2105 init_on_alloc= [MM,EARLY] Fill newly allocated pages and heap objects with
2108 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
2110 init_on_free= [MM,EARLY] Fill freed pages and heap objects with zeroes.
2112 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
2114 init_pkru= [X86] Specify the default memory protection keys rights
2115 register contents for all processes. 0x55555554 by
2116 default (disallow access to all but pkey 0). Can
2117 override in debugfs after boot.
2119 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
2122 int_pln_enable [X86] Enable power limit notification interrupt
2124 integrity_audit=[IMA]
2125 Format: { "0" | "1" }
2126 0 -- basic integrity auditing messages. (Default)
2127 1 -- additional integrity auditing messages.
2129 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
2131 Enable intel iommu driver.
2133 Disable intel iommu driver.
2134 igfx_off [Default Off]
2135 By default, gfx is mapped as normal device. If a gfx
2136 device has a dedicated DMAR unit, the DMAR unit is
2137 bypassed by not enabling DMAR with this option. In
2138 this case, gfx device will use physical address for
2140 strict [Default Off]
2141 Deprecated, equivalent to iommu.strict=1.
2142 sp_off [Default Off]
2143 By default, super page will be supported if Intel IOMMU
2144 has the capability. With this option, super page will
2147 Enable the Intel IOMMU scalable mode if the hardware
2148 advertises that it has support for the scalable mode
2151 Disallow use of the Intel IOMMU scalable mode.
2152 tboot_noforce [Default Off]
2153 Do not force the Intel IOMMU enabled under tboot.
2154 By default, tboot will force Intel IOMMU on, which
2155 could harm performance of some high-throughput
2156 devices like 40GBit network cards, even if identity
2158 Note that using this option lowers the security
2159 provided by tboot because it makes the system
2160 vulnerable to DMA attacks.
2162 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
2163 0 disables intel_idle and fall back on acpi_idle.
2164 1 to 9 specify maximum depth of C-state.
2166 intel_pstate= [X86,EARLY]
2168 Do not enable intel_pstate as the default
2169 scaling driver for the supported processors
2171 Use intel_pstate driver to bypass the scaling
2172 governors layer of cpufreq and provides it own
2173 algorithms for p-state selection. There are two
2174 P-state selection algorithms provided by
2175 intel_pstate in the active mode: powersave and
2176 performance. The way they both operate depends
2177 on whether or not the hardware managed P-states
2178 (HWP) feature has been enabled in the processor
2179 and possibly on the processor model.
2181 Use intel_pstate as a scaling driver, but configure it
2182 to work with generic cpufreq governors (instead of
2183 enabling its internal governor). This mode cannot be
2184 used along with the hardware-managed P-states (HWP)
2187 Enable intel_pstate on systems that prohibit it by default
2188 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2189 instead of acpi-cpufreq may disable platform features, such
2190 as thermal controls and power capping, that rely on ACPI
2191 P-States information being indicated to OSPM and therefore
2192 should be used with caution. This option does not work with
2193 processors that aren't supported by the intel_pstate driver
2194 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2196 Do not enable hardware P state control (HWP)
2199 Only load intel_pstate on systems which support
2200 hardware P state control (HWP) if available.
2202 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2203 Description Table, specifies preferred power management
2204 profile as "Enterprise Server" or "Performance Server",
2205 then this feature is turned on by default.
2207 Allow per-logical-CPU P-State performance control limits using
2208 cpufreq sysfs interface
2210 intremap= [X86-64,Intel-IOMMU,EARLY]
2211 on enable Interrupt Remapping (default)
2212 off disable Interrupt Remapping
2213 nosid disable Source ID checking
2215 BIOS x2APIC opt-out request will be ignored
2216 nopost disable Interrupt Posting
2218 iomem= Disable strict checking of access to MMIO memory
2219 strict regions from userspace.
2234 nobypass [PPC/POWERNV]
2235 Disable IOMMU bypass, using IOMMU for PCI devices.
2237 iommu.forcedac= [ARM64,X86,EARLY] Control IOVA allocation for PCI devices.
2238 Format: { "0" | "1" }
2239 0 - Try to allocate a 32-bit DMA address first, before
2240 falling back to the full range if needed.
2241 1 - Allocate directly from the full usable range,
2242 forcing Dual Address Cycle for PCI cards supporting
2243 greater than 32-bit addressing.
2245 iommu.strict= [ARM64,X86,S390,EARLY] Configure TLB invalidation behaviour
2246 Format: { "0" | "1" }
2248 Request that DMA unmap operations use deferred
2249 invalidation of hardware TLBs, for increased
2250 throughput at the cost of reduced device isolation.
2251 Will fall back to strict mode if not supported by
2252 the relevant IOMMU driver.
2254 DMA unmap operations invalidate IOMMU hardware TLBs
2256 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2257 Note: on x86, strict mode specified via one of the
2258 legacy driver-specific options takes precedence.
2261 [ARM64,X86,EARLY] Configure DMA to bypass the IOMMU by default.
2262 Format: { "0" | "1" }
2263 0 - Use IOMMU translation for DMA.
2264 1 - Bypass the IOMMU for DMA.
2265 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2267 io7= [HW] IO7 for Marvel-based Alpha systems
2268 See comment before marvel_specify_io7 in
2269 arch/alpha/kernel/core_marvel.c.
2271 io_delay= [X86,EARLY] I/O delay method
2273 Standard port 0x80 based delay
2275 Alternate port 0xed based delay (needed on some systems)
2277 Simple two microseconds delay
2282 See Documentation/admin-guide/nfs/nfsroot.rst.
2284 ipcmni_extend [KNL,EARLY] Extend the maximum number of unique System V
2285 IPC identifiers from 32,768 to 16,777,216.
2287 irqaffinity= [SMP] Set the default irq affinity mask
2288 The argument is a cpu list, as described above.
2290 irqchip.gicv2_force_probe=
2293 Force the kernel to look for the second 4kB page
2294 of a GICv2 controller even if the memory range
2295 exposed by the device tree is too small.
2297 irqchip.gicv3_nolpi=
2299 Force the kernel to ignore the availability of
2300 LPIs (and by consequence ITSs). Intended for system
2301 that use the kernel as a bootloader, and thus want
2302 to let secondary kernels in charge of setting up
2305 irqchip.gicv3_pseudo_nmi= [ARM64,EARLY]
2306 Enables support for pseudo-NMIs in the kernel. This
2307 requires the kernel to be built with
2308 CONFIG_ARM64_PSEUDO_NMI.
2311 When an interrupt is not handled search all handlers
2312 for it. Intended to get systems with badly broken
2316 When an interrupt is not handled search all handlers
2317 for it. Also check all handlers each timer
2318 interrupt. Intended to get systems with badly broken
2322 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2324 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2325 [Deprecated - use cpusets instead]
2326 Format: [flag-list,]<cpu-list>
2328 Specify one or more CPUs to isolate from disturbances
2329 specified in the flag list (default: domain):
2332 Disable the tick when a single task runs.
2334 A residual 1Hz tick is offloaded to workqueues, which you
2335 need to affine to housekeeping through the global
2336 workqueue's affinity configured via the
2337 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2338 by using the 'domain' flag described below.
2340 NOTE: by default the global workqueue runs on all CPUs,
2341 so to protect individual CPUs the 'cpumask' file has to
2342 be configured manually after bootup.
2345 Isolate from the general SMP balancing and scheduling
2346 algorithms. Note that performing domain isolation this way
2347 is irreversible: it's not possible to bring back a CPU to
2348 the domains once isolated through isolcpus. It's strongly
2349 advised to use cpusets instead to disable scheduler load
2350 balancing through the "cpuset.sched_load_balance" file.
2351 It offers a much more flexible interface where CPUs can
2352 move in and out of an isolated set anytime.
2354 You can move a process onto or off an "isolated" CPU via
2355 the CPU affinity syscalls or cpuset.
2356 <cpu number> begins at 0 and the maximum value is
2357 "number of CPUs in system - 1".
2361 Isolate from being targeted by managed interrupts
2362 which have an interrupt mask containing isolated
2363 CPUs. The affinity of managed interrupts is
2364 handled by the kernel and cannot be changed via
2365 the /proc/irq/* interfaces.
2367 This isolation is best effort and only effective
2368 if the automatically assigned interrupt mask of a
2369 device queue contains isolated and housekeeping
2370 CPUs. If housekeeping CPUs are online then such
2371 interrupts are directed to the housekeeping CPU
2372 so that IO submitted on the housekeeping CPU
2373 cannot disturb the isolated CPU.
2375 If a queue's affinity mask contains only isolated
2376 CPUs then this parameter has no effect on the
2377 interrupt routing decision, though interrupts are
2378 only delivered when tasks running on those
2379 isolated CPUs submit IO. IO submitted on
2380 housekeeping CPUs has no influence on those
2383 The format of <cpu-list> is described above.
2387 ivrs_ioapic [HW,X86-64]
2388 Provide an override to the IOAPIC-ID<->DEVICE-ID
2389 mapping provided in the IVRS ACPI table.
2390 By default, PCI segment is 0, and can be omitted.
2392 For example, to map IOAPIC-ID decimal 10 to
2393 PCI segment 0x1 and PCI device 00:14.0,
2394 write the parameter as:
2395 ivrs_ioapic=10@0001:00:14.0
2398 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0
2399 write the parameter as:
2400 ivrs_ioapic[10]=00:14.0
2401 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and
2402 PCI device 00:14.0 write the parameter as:
2403 ivrs_ioapic[10]=0001:00:14.0
2405 ivrs_hpet [HW,X86-64]
2406 Provide an override to the HPET-ID<->DEVICE-ID
2407 mapping provided in the IVRS ACPI table.
2408 By default, PCI segment is 0, and can be omitted.
2410 For example, to map HPET-ID decimal 10 to
2411 PCI segment 0x1 and PCI device 00:14.0,
2412 write the parameter as:
2413 ivrs_hpet=10@0001:00:14.0
2416 * To map HPET-ID decimal 0 to PCI device 00:14.0
2417 write the parameter as:
2418 ivrs_hpet[0]=00:14.0
2419 * To map HPET-ID decimal 10 to PCI segment 0x1 and
2420 PCI device 00:14.0 write the parameter as:
2421 ivrs_ioapic[10]=0001:00:14.0
2423 ivrs_acpihid [HW,X86-64]
2424 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2425 mapping provided in the IVRS ACPI table.
2426 By default, PCI segment is 0, and can be omitted.
2428 For example, to map UART-HID:UID AMD0020:0 to
2429 PCI segment 0x1 and PCI device ID 00:14.5,
2430 write the parameter as:
2431 ivrs_acpihid=AMD0020:0@0001:00:14.5
2434 * To map UART-HID:UID AMD0020:0 to PCI segment is 0,
2435 PCI device ID 00:14.5, write the parameter as:
2436 ivrs_acpihid[00:14.5]=AMD0020:0
2437 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and
2438 PCI device ID 00:14.5, write the parameter as:
2439 ivrs_acpihid[0001:00:14.5]=AMD0020:0
2441 js= [HW,JOY] Analog joystick
2442 See Documentation/input/joydev/joystick.rst.
2445 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2446 report on every invalid memory access. Without this
2447 parameter KASAN will print report only for the first
2450 keep_bootcon [KNL,EARLY]
2451 Do not unregister boot console at start. This is only
2452 useful for debugging when something happens in the window
2453 between unregistering the boot console and initializing
2456 keepinitrd [HW,ARM] See retain_initrd.
2458 kernelcore= [KNL,X86,IA-64,PPC,EARLY]
2459 Format: nn[KMGTPE] | nn% | "mirror"
2460 This parameter specifies the amount of memory usable by
2461 the kernel for non-movable allocations. The requested
2462 amount is spread evenly throughout all nodes in the
2463 system as ZONE_NORMAL. The remaining memory is used for
2464 movable memory in its own zone, ZONE_MOVABLE. In the
2465 event, a node is too small to have both ZONE_NORMAL and
2466 ZONE_MOVABLE, kernelcore memory will take priority and
2467 other nodes will have a larger ZONE_MOVABLE.
2469 ZONE_MOVABLE is used for the allocation of pages that
2470 may be reclaimed or moved by the page migration
2471 subsystem. Note that allocations like PTEs-from-HighMem
2472 still use the HighMem zone if it exists, and the Normal
2473 zone if it does not.
2475 It is possible to specify the exact amount of memory in
2476 the form of "nn[KMGTPE]", a percentage of total system
2477 memory in the form of "nn%", or "mirror". If "mirror"
2478 option is specified, mirrored (reliable) memory is used
2479 for non-movable allocations and remaining memory is used
2480 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2481 are exclusive, so you cannot specify multiple forms.
2483 kgdbdbgp= [KGDB,HW,EARLY] kgdb over EHCI usb debug port.
2484 Format: <Controller#>[,poll interval]
2485 The controller # is the number of the ehci usb debug
2486 port as it is probed via PCI. The poll interval is
2487 optional and is the number seconds in between
2488 each poll cycle to the debug port in case you need
2489 the functionality for interrupting the kernel with
2490 gdb or control-c on the dbgp connection. When
2491 not using this parameter you use sysrq-g to break into
2492 the kernel debugger.
2494 kgdboc= [KGDB,HW] kgdb over consoles.
2495 Requires a tty driver that supports console polling,
2496 or a supported polling keyboard driver (non-usb).
2497 Serial only format: <serial_device>[,baud]
2498 keyboard only format: kbd
2499 keyboard and serial format: kbd,<serial_device>[,baud]
2500 Optional Kernel mode setting:
2501 kms, kbd format: kms,kbd
2502 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2504 kgdboc_earlycon= [KGDB,HW,EARLY]
2505 If the boot console provides the ability to read
2506 characters and can work in polling mode, you can use
2507 this parameter to tell kgdb to use it as a backend
2508 until the normal console is registered. Intended to
2509 be used together with the kgdboc parameter which
2510 specifies the normal console to transition to.
2512 The name of the early console should be specified
2513 as the value of this parameter. Note that the name of
2514 the early console might be different than the tty
2515 name passed to kgdboc. It's OK to leave the value
2516 blank and the first boot console that implements
2517 read() will be picked.
2519 kgdbwait [KGDB,EARLY] Stop kernel execution and enter the
2520 kernel debugger at the earliest opportunity.
2522 kmac= [MIPS] Korina ethernet MAC address.
2523 Configure the RouterBoard 532 series on-chip
2524 Ethernet adapter MAC address.
2526 kmemleak= [KNL,EARLY] Boot-time kmemleak enable/disable
2527 Valid arguments: on, off
2529 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2532 kprobe_event=[probe-list]
2533 [FTRACE] Add kprobe events and enable at boot time.
2534 The probe-list is a semicolon delimited list of probe
2535 definitions. Each definition is same as kprobe_events
2536 interface, but the parameters are comma delimited.
2537 For example, to add a kprobe event on vfs_read with
2538 arg1 and arg2, add to the command line;
2540 kprobe_event=p,vfs_read,$arg1,$arg2
2542 See also Documentation/trace/kprobetrace.rst "Kernel
2543 Boot Parameter" section.
2545 kpti= [ARM64,EARLY] Control page table isolation of
2546 user and kernel address spaces.
2547 Default: enabled on cores which need mitigation.
2551 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires
2552 CONFIG_KUNIT to be set to be fully enabled. The
2553 default value can be overridden via
2554 KUNIT_DEFAULT_ENABLED.
2555 Default is 1 (enabled)
2557 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2558 Default is 0 (don't ignore, but inject #GP)
2560 kvm.eager_page_split=
2561 [KVM,X86] Controls whether or not KVM will try to
2562 proactively split all huge pages during dirty logging.
2563 Eager page splitting reduces interruptions to vCPU
2564 execution by eliminating the write-protection faults
2565 and MMU lock contention that would otherwise be
2566 required to split huge pages lazily.
2568 VM workloads that rarely perform writes or that write
2569 only to a small region of VM memory may benefit from
2570 disabling eager page splitting to allow huge pages to
2571 still be used for reads.
2573 The behavior of eager page splitting depends on whether
2574 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If
2575 disabled, all huge pages in a memslot will be eagerly
2576 split when dirty logging is enabled on that memslot. If
2577 enabled, eager page splitting will be performed during
2578 the KVM_CLEAR_DIRTY ioctl, and only for the pages being
2581 Eager page splitting is only supported when kvm.tdp_mmu=Y.
2585 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2586 Default is false (don't support).
2589 [KVM] Controls the software workaround for the
2590 X86_BUG_ITLB_MULTIHIT bug.
2591 force : Always deploy workaround.
2592 off : Never deploy workaround.
2593 auto : Deploy workaround based on the presence of
2594 X86_BUG_ITLB_MULTIHIT.
2598 If the software workaround is enabled for the host,
2599 guests do need not to enable it for nested guests.
2601 kvm.nx_huge_pages_recovery_ratio=
2602 [KVM] Controls how many 4KiB pages are periodically zapped
2603 back to huge pages. 0 disables the recovery, otherwise if
2604 the value is N KVM will zap 1/Nth of the 4KiB pages every
2605 period (see below). The default is 60.
2607 kvm.nx_huge_pages_recovery_period_ms=
2608 [KVM] Controls the time period at which KVM zaps 4KiB pages
2609 back to huge pages. If the value is a non-zero N, KVM will
2610 zap a portion (see ratio above) of the pages every N msecs.
2611 If the value is 0 (the default), KVM will pick a period based
2612 on the ratio, such that a page is zapped after 1 hour on average.
2614 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in
2615 KVM/SVM. Default is 1 (enabled).
2617 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables,
2618 a.k.a. Two-Dimensional Page Tables. Default is 1
2619 (enabled). Disable by KVM if hardware lacks support
2623 [KVM,ARM,EARLY] Select one of KVM/arm64's modes of
2626 none: Forcefully disable KVM.
2628 nvhe: Standard nVHE-based mode, without support for
2631 protected: nVHE-based mode with support for guests whose
2632 state is kept private from the host.
2634 nested: VHE-based mode with support for nested
2635 virtualization. Requires at least ARMv8.3
2638 Defaults to VHE/nVHE based on hardware support. Setting
2639 mode to "protected" will disable kexec and hibernation
2640 for the host. "nested" is experimental and should be
2641 used with extreme caution.
2643 kvm-arm.vgic_v3_group0_trap=
2644 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-0
2647 kvm-arm.vgic_v3_group1_trap=
2648 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-1
2651 kvm-arm.vgic_v3_common_trap=
2652 [KVM,ARM,EARLY] Trap guest accesses to GICv3 common
2655 kvm-arm.vgic_v4_enable=
2656 [KVM,ARM,EARLY] Allow use of GICv4 for direct
2659 kvm_cma_resv_ratio=n [PPC,EARLY]
2660 Reserves given percentage from system memory area for
2661 contiguous memory allocation for KVM hash pagetable
2663 By default it reserves 5% of total system memory.
2667 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables,
2668 a.k.a. Two-Dimensional Page Tables. Default is 1
2669 (enabled). Disable by KVM if hardware lacks support
2672 kvm-intel.emulate_invalid_guest_state=
2673 [KVM,Intel] Control whether to emulate invalid guest
2674 state. Ignored if kvm-intel.enable_unrestricted_guest=1,
2675 as guest state is never invalid for unrestricted
2676 guests. This param doesn't apply to nested guests (L2),
2677 as KVM never emulates invalid L2 guest state.
2678 Default is 1 (enabled).
2680 kvm-intel.flexpriority=
2681 [KVM,Intel] Control KVM's use of FlexPriority feature
2682 (TPR shadow). Default is 1 (enabled). Disable by KVM if
2683 hardware lacks support for it.
2686 [KVM,Intel] Control nested virtualization feature in
2687 KVM/VMX. Default is 1 (enabled).
2689 kvm-intel.unrestricted_guest=
2690 [KVM,Intel] Control KVM's use of unrestricted guest
2691 feature (virtualized real and unpaged mode). Default
2692 is 1 (enabled). Disable by KVM if EPT is disabled or
2693 hardware lacks support for it.
2695 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2698 Valid arguments: never, cond, always
2700 always: L1D cache flush on every VMENTER.
2701 cond: Flush L1D on VMENTER only when the code between
2702 VMEXIT and VMENTER can leak host memory.
2703 never: Disables the mitigation
2705 Default is cond (do L1 cache flush in specific instances)
2707 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor
2708 Identification feature (tagged TLBs). Default is 1
2709 (enabled). Disable by KVM if hardware lacks support
2712 l1d_flush= [X86,INTEL,EARLY]
2713 Control mitigation for L1D based snooping vulnerability.
2715 Certain CPUs are vulnerable to an exploit against CPU
2716 internal buffers which can forward information to a
2717 disclosure gadget under certain conditions.
2719 In vulnerable processors, the speculatively
2720 forwarded data can be used in a cache side channel
2721 attack, to access data to which the attacker does
2722 not have direct access.
2724 This parameter controls the mitigation. The
2727 on - enable the interface for the mitigation
2729 l1tf= [X86,EARLY] Control mitigation of the L1TF vulnerability on
2732 The kernel PTE inversion protection is unconditionally
2733 enabled and cannot be disabled.
2736 Provides all available mitigations for the
2737 L1TF vulnerability. Disables SMT and
2738 enables all mitigations in the
2739 hypervisors, i.e. unconditional L1D flush.
2741 SMT control and L1D flush control via the
2742 sysfs interface is still possible after
2743 boot. Hypervisors will issue a warning
2744 when the first VM is started in a
2745 potentially insecure configuration,
2746 i.e. SMT enabled or L1D flush disabled.
2749 Same as 'full', but disables SMT and L1D
2750 flush runtime control. Implies the
2751 'nosmt=force' command line option.
2752 (i.e. sysfs control of SMT is disabled.)
2755 Leaves SMT enabled and enables the default
2756 hypervisor mitigation, i.e. conditional
2759 SMT control and L1D flush control via the
2760 sysfs interface is still possible after
2761 boot. Hypervisors will issue a warning
2762 when the first VM is started in a
2763 potentially insecure configuration,
2764 i.e. SMT enabled or L1D flush disabled.
2768 Disables SMT and enables the default
2769 hypervisor mitigation.
2771 SMT control and L1D flush control via the
2772 sysfs interface is still possible after
2773 boot. Hypervisors will issue a warning
2774 when the first VM is started in a
2775 potentially insecure configuration,
2776 i.e. SMT enabled or L1D flush disabled.
2779 Same as 'flush', but hypervisors will not
2780 warn when a VM is started in a potentially
2781 insecure configuration.
2784 Disables hypervisor mitigations and doesn't
2786 It also drops the swap size and available
2787 RAM limit restriction on both hypervisor and
2792 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2798 lapic [X86-32,APIC,EARLY] Enable the local APIC even if BIOS
2801 lapic= [X86,APIC] Do not use TSC deadline
2802 value for LAPIC timer one-shot implementation. Default
2803 back to the programmable timer unit in the LAPIC.
2804 Format: notscdeadline
2806 lapic_timer_c2_ok [X86,APIC,EARLY] trust the local apic timer
2809 libata.dma= [LIBATA] DMA control
2810 libata.dma=0 Disable all PATA and SATA DMA
2811 libata.dma=1 PATA and SATA Disk DMA only
2812 libata.dma=2 ATAPI (CDROM) DMA only
2813 libata.dma=4 Compact Flash DMA only
2814 Combinations also work, so libata.dma=3 enables DMA
2815 for disks and CDROMs, but not CFs.
2817 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2818 libata.ignore_hpa=0 keep BIOS limits (default)
2819 libata.ignore_hpa=1 ignore limits, using full disk
2821 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2825 libata.force= [LIBATA] Force configurations. The format is a comma-
2826 separated list of "[ID:]VAL" where ID is PORT[.DEVICE].
2827 PORT and DEVICE are decimal numbers matching port, link
2828 or device. Basically, it matches the ATA ID string
2829 printed on console by libata. If the whole ID part is
2830 omitted, the last PORT and DEVICE values are used. If
2831 ID hasn't been specified yet, the configuration applies
2832 to all ports, links and devices.
2834 If only DEVICE is omitted, the parameter applies to
2835 the port and all links and devices behind it. DEVICE
2836 number of 0 either selects the first device or the
2837 first fan-out link behind PMP device. It does not
2838 select the host link. DEVICE number of 15 selects the
2839 host link and device attached to it.
2841 The VAL specifies the configuration to force. As long
2842 as there is no ambiguity, shortcut notation is allowed.
2843 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2844 The following configurations can be forced.
2846 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2847 Any ID with matching PORT is used.
2849 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2851 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2852 udma[/][16,25,33,44,66,100,133] notation is also
2855 * nohrst, nosrst, norst: suppress hard, soft and both
2858 * rstonce: only attempt one reset during hot-unplug
2861 * [no]dbdelay: Enable or disable the extra 200ms delay
2862 before debouncing a link PHY and device presence
2865 * [no]ncq: Turn on or off NCQ.
2867 * [no]ncqtrim: Enable or disable queued DSM TRIM.
2869 * [no]ncqati: Enable or disable NCQ trim on ATI chipset.
2871 * [no]trim: Enable or disable (unqueued) TRIM.
2873 * trim_zero: Indicate that TRIM command zeroes data.
2875 * max_trim_128m: Set 128M maximum trim size limit.
2877 * [no]dma: Turn on or off DMA transfers.
2879 * atapi_dmadir: Enable ATAPI DMADIR bridge support.
2881 * atapi_mod16_dma: Enable the use of ATAPI DMA for
2882 commands that are not a multiple of 16 bytes.
2884 * [no]dmalog: Enable or disable the use of the
2885 READ LOG DMA EXT command to access logs.
2887 * [no]iddevlog: Enable or disable access to the
2888 identify device data log.
2890 * [no]logdir: Enable or disable access to the general
2891 purpose log directory.
2893 * max_sec_128: Set transfer size limit to 128 sectors.
2895 * max_sec_1024: Set or clear transfer size limit to
2898 * max_sec_lba48: Set or clear transfer size limit to
2901 * [no]lpm: Enable or disable link power management.
2903 * [no]setxfer: Indicate if transfer speed mode setting
2906 * [no]fua: Disable or enable FUA (Force Unit Access)
2907 support for devices supporting this feature.
2909 * dump_id: Dump IDENTIFY data.
2911 * disable: Disable this device.
2913 If there are multiple matching configurations changing
2914 the same attribute, the last one is used.
2916 load_ramdisk= [RAM] [Deprecated]
2918 lockd.nlm_grace_period=P [NFS] Assign grace period.
2921 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2924 lockd.nlm_timeout=T [NFS] Assign timeout value.
2927 lockd.nlm_udpport=M [NFS] Assign UDP port.
2930 lockdown= [SECURITY,EARLY]
2931 { integrity | confidentiality }
2932 Enable the kernel lockdown feature. If set to
2933 integrity, kernel features that allow userland to
2934 modify the running kernel are disabled. If set to
2935 confidentiality, kernel features that allow userland
2936 to extract confidential information from the kernel
2939 locktorture.acq_writer_lim= [KNL]
2940 Set the time limit in jiffies for a lock
2941 acquisition. Acquisitions exceeding this limit
2942 will result in a splat once they do complete.
2944 locktorture.bind_readers= [KNL]
2945 Specify the list of CPUs to which the readers are
2948 locktorture.bind_writers= [KNL]
2949 Specify the list of CPUs to which the writers are
2952 locktorture.call_rcu_chains= [KNL]
2953 Specify the number of self-propagating call_rcu()
2954 chains to set up. These are used to ensure that
2955 there is a high probability of an RCU grace period
2956 in progress at any given time. Defaults to 0,
2957 which disables these call_rcu() chains.
2959 locktorture.long_hold= [KNL]
2960 Specify the duration in milliseconds for the
2961 occasional long-duration lock hold time. Defaults
2962 to 100 milliseconds. Select 0 to disable.
2964 locktorture.nested_locks= [KNL]
2965 Specify the maximum lock nesting depth that
2966 locktorture is to exercise, up to a limit of 8
2967 (MAX_NESTED_LOCKS). Specify zero to disable.
2968 Note that this parameter is ineffective on types
2969 of locks that do not support nested acquisition.
2971 locktorture.nreaders_stress= [KNL]
2972 Set the number of locking read-acquisition kthreads.
2973 Defaults to being automatically set based on the
2974 number of online CPUs.
2976 locktorture.nwriters_stress= [KNL]
2977 Set the number of locking write-acquisition kthreads.
2979 locktorture.onoff_holdoff= [KNL]
2980 Set time (s) after boot for CPU-hotplug testing.
2982 locktorture.onoff_interval= [KNL]
2983 Set time (s) between CPU-hotplug operations, or
2984 zero to disable CPU-hotplug testing.
2986 locktorture.rt_boost= [KNL]
2987 Do periodic testing of real-time lock priority
2988 boosting. Select 0 to disable, 1 to boost
2989 only rt_mutex, and 2 to boost unconditionally.
2990 Defaults to 2, which might seem to be an
2991 odd choice, but which should be harmless for
2992 non-real-time spinlocks, due to their disabling
2993 of preemption. Note that non-realtime mutexes
2996 locktorture.rt_boost_factor= [KNL]
2997 Number that determines how often and for how
2998 long priority boosting is exercised. This is
2999 scaled down by the number of writers, so that the
3000 number of boosts per unit time remains roughly
3001 constant as the number of writers increases.
3002 On the other hand, the duration of each boost
3003 increases with the number of writers.
3005 locktorture.shuffle_interval= [KNL]
3006 Set task-shuffle interval (jiffies). Shuffling
3007 tasks allows some CPUs to go into dyntick-idle
3008 mode during the locktorture test.
3010 locktorture.shutdown_secs= [KNL]
3011 Set time (s) after boot system shutdown. This
3012 is useful for hands-off automated testing.
3014 locktorture.stat_interval= [KNL]
3015 Time (s) between statistics printk()s.
3017 locktorture.stutter= [KNL]
3018 Time (s) to stutter testing, for example,
3019 specifying five seconds causes the test to run for
3020 five seconds, wait for five seconds, and so on.
3021 This tests the locking primitive's ability to
3022 transition abruptly to and from idle.
3024 locktorture.torture_type= [KNL]
3025 Specify the locking implementation to test.
3027 locktorture.verbose= [KNL]
3028 Enable additional printk() statements.
3030 locktorture.writer_fifo= [KNL]
3031 Run the write-side locktorture kthreads at
3032 sched_set_fifo() real-time priority.
3034 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
3037 loglevel= [KNL,EARLY]
3038 All Kernel Messages with a loglevel smaller than the
3039 console loglevel will be printed to the console. It can
3040 also be changed with klogd or other programs. The
3041 loglevels are defined as follows:
3043 0 (KERN_EMERG) system is unusable
3044 1 (KERN_ALERT) action must be taken immediately
3045 2 (KERN_CRIT) critical conditions
3046 3 (KERN_ERR) error conditions
3047 4 (KERN_WARNING) warning conditions
3048 5 (KERN_NOTICE) normal but significant condition
3049 6 (KERN_INFO) informational
3050 7 (KERN_DEBUG) debug-level messages
3052 log_buf_len=n[KMG] [KNL,EARLY]
3053 Sets the size of the printk ring buffer, in bytes.
3054 n must be a power of two and greater than the
3055 minimal size. The minimal size is defined by
3056 LOG_BUF_SHIFT kernel config parameter. There
3057 is also CONFIG_LOG_CPU_MAX_BUF_SHIFT config
3058 parameter that allows to increase the default size
3059 depending on the number of CPUs. See init/Kconfig
3062 logo.nologo [FB] Disables display of the built-in Linux logo.
3063 This may be used to provide more screen space for
3064 kernel log messages and is useful when debugging
3065 kernel boot problems.
3067 lp=0 [LP] Specify parallel ports to use, e.g,
3068 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
3069 lp=reset first parallel port). 'lp=0' disables the
3070 lp=auto printer driver. 'lp=reset' (which can be
3071 specified in addition to the ports) causes
3072 attached printers to be reset. Using
3073 lp=port1,port2,... specifies the parallel ports
3074 to associate lp devices with, starting with
3075 lp0. A port specification may be 'none' to skip
3076 that lp device, or a parport name such as
3077 'parport0'. Specifying 'lp=auto' instead of a
3078 port specification list means that device IDs
3079 from each port should be examined, to see if
3080 an IEEE 1284-compliant printer is attached; if
3081 so, the driver will manage that printer.
3082 See also header of drivers/char/lp.c.
3085 Sets loops_per_jiffy to given constant, thus avoiding
3086 time-consuming boot-time autodetection (up to 250 ms per
3087 CPU). 0 enables autodetection (default). To determine
3088 the correct value for your kernel, boot with normal
3089 autodetection and see what value is printed. Note that
3090 on SMP systems the preset will be applied to all CPUs,
3091 which is likely to cause problems if your CPUs need
3092 significantly divergent settings. An incorrect value
3093 will cause delays in the kernel to be wrong, leading to
3094 unpredictable I/O errors and other breakage. Although
3095 unlikely, in the extreme case this might damage your
3099 Format: <io>,<irq>,<dma>
3101 lsm.debug [SECURITY] Enable LSM initialization debugging output.
3104 [SECURITY] Choose order of LSM initialization. This
3105 overrides CONFIG_LSM, and the "security=" parameter.
3107 machvec= [IA-64] Force the use of a particular machine-vector
3108 (machvec) in a generic kernel.
3109 Example: machvec=hpzx1
3111 machtype= [Loongson] Share the same kernel image file between
3112 different yeeloong laptops.
3113 Example: machtype=lemote-yeeloong-2f-7inch
3115 max_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory greater
3116 than or equal to this physical address is ignored.
3118 maxcpus= [SMP,EARLY] Maximum number of processors that an SMP kernel
3119 will bring up during bootup. maxcpus=n : n >= 0 limits
3120 the kernel to bring up 'n' processors. Surely after
3121 bootup you can bring up the other plugged cpu by executing
3122 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
3123 only takes effect during system bootup.
3124 While n=0 is a special case, it is equivalent to "nosmp",
3125 which also disables the IO APIC.
3127 max_loop= [LOOP] The number of loop block devices that get
3128 (loop.max_loop) unconditionally pre-created at init time. The default
3129 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
3130 of statically allocating a predefined number, loop
3131 devices can be requested on-demand with the
3132 /dev/loop-control interface.
3134 mce [X86-32] Machine Check Exception
3136 mce=option [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
3138 md= [HW] RAID subsystems devices and level
3139 See Documentation/admin-guide/md.rst.
3142 Format: <first>,<last>
3143 Specifies range of consoles to be captured by the MDA.
3145 mds= [X86,INTEL,EARLY]
3146 Control mitigation for the Micro-architectural Data
3147 Sampling (MDS) vulnerability.
3149 Certain CPUs are vulnerable to an exploit against CPU
3150 internal buffers which can forward information to a
3151 disclosure gadget under certain conditions.
3153 In vulnerable processors, the speculatively
3154 forwarded data can be used in a cache side channel
3155 attack, to access data to which the attacker does
3156 not have direct access.
3158 This parameter controls the MDS mitigation. The
3161 full - Enable MDS mitigation on vulnerable CPUs
3162 full,nosmt - Enable MDS mitigation and disable
3163 SMT on vulnerable CPUs
3164 off - Unconditionally disable MDS mitigation
3166 On TAA-affected machines, mds=off can be prevented by
3167 an active TAA mitigation as both vulnerabilities are
3168 mitigated with the same mechanism so in order to disable
3169 this mitigation, you need to specify tsx_async_abort=off
3172 Not specifying this option is equivalent to
3175 For details see: Documentation/admin-guide/hw-vuln/mds.rst
3177 mem=nn[KMG] [HEXAGON,EARLY] Set the memory size.
3178 Must be specified, otherwise memory size will be 0.
3180 mem=nn[KMG] [KNL,BOOT,EARLY] Force usage of a specific amount
3181 of memory Amount of memory to be used in cases
3185 2 when the kernel is not able to see the whole system memory;
3186 3 memory that lies after 'mem=' boundary is excluded from
3187 the hypervisor, then assigned to KVM guests.
3188 4 to limit the memory available for kdump kernel.
3190 [ARC,MICROBLAZE] - the limit applies only to low memory,
3191 high memory is not affected.
3193 [ARM64] - only limits memory covered by the linear
3194 mapping. The NOMAP regions are not affected.
3196 [X86] Work as limiting max address. Use together
3197 with memmap= to avoid physical address space collisions.
3198 Without memmap= PCI devices could be placed at addresses
3199 belonging to unused RAM.
3201 Note that this only takes effects during boot time since
3202 in above case 3, memory may need be hot added after boot
3203 if system memory of hypervisor is not sufficient.
3206 [ARM,MIPS,EARLY] - override the memory layout
3207 reported by firmware.
3208 Define a memory region of size nn[KMG] starting at
3210 Multiple different regions can be specified with
3211 multiple mem= parameters on the command line.
3213 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
3216 memblock=debug [KNL,EARLY] Enable memblock debug messages.
3219 [KNL,SH] Allow user to override the default size for
3220 per-device physically contiguous DMA buffers.
3222 memhp_default_state=online/offline/online_kernel/online_movable
3223 [KNL] Set the initial state for the memory hotplug
3224 onlining policy. If not specified, the default value is
3225 set according to the
3226 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
3228 See Documentation/admin-guide/mm/memory-hotplug.rst.
3230 memmap=exactmap [KNL,X86,EARLY] Enable setting of an exact
3231 E820 memory map, as specified by the user.
3232 Such memmap=exactmap lines can be constructed based on
3233 BIOS output or other requirements. See the memmap=nn@ss
3236 memmap=nn[KMG]@ss[KMG]
3237 [KNL, X86,MIPS,XTENSA,EARLY] Force usage of a specific region of memory.
3238 Region of memory to be used is from ss to ss+nn.
3239 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
3240 which limits max address to nn[KMG].
3241 Multiple different regions can be specified,
3244 memmap=100M@2G,100M#3G,1G!1024G
3246 memmap=nn[KMG]#ss[KMG]
3247 [KNL,ACPI,EARLY] Mark specific memory as ACPI data.
3248 Region of memory to be marked is from ss to ss+nn.
3250 memmap=nn[KMG]$ss[KMG]
3251 [KNL,ACPI,EARLY] Mark specific memory as reserved.
3252 Region of memory to be reserved is from ss to ss+nn.
3253 Example: Exclude memory from 0x18690000-0x1869ffff
3254 memmap=64K$0x18690000
3256 memmap=0x10000$0x18690000
3257 Some bootloaders may need an escape character before '$',
3258 like Grub2, otherwise '$' and the following number
3261 memmap=nn[KMG]!ss[KMG,EARLY]
3262 [KNL,X86] Mark specific memory as protected.
3263 Region of memory to be used, from ss to ss+nn.
3264 The memory region may be marked as e820 type 12 (0xc)
3265 and is NVDIMM or ADR memory.
3267 memmap=<size>%<offset>-<oldtype>+<newtype>
3268 [KNL,ACPI,EARLY] Convert memory within the specified region
3269 from <oldtype> to <newtype>. If "-<oldtype>" is left
3270 out, the whole region will be marked as <newtype>,
3271 even if previously unavailable. If "+<newtype>" is left
3272 out, matching memory will be removed. Types are
3273 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
3274 3 = ACPI, 12 = PRAM.
3276 memory_corruption_check=0/1 [X86,EARLY]
3277 Some BIOSes seem to corrupt the first 64k of
3278 memory when doing things like suspend/resume.
3279 Setting this option will scan the memory
3280 looking for corruption. Enabling this will
3281 both detect corruption and prevent the kernel
3282 from using the memory being corrupted.
3283 However, its intended as a diagnostic tool; if
3284 repeatable BIOS-originated corruption always
3285 affects the same memory, you can use memmap=
3286 to prevent the kernel from using that memory.
3288 memory_corruption_check_size=size [X86,EARLY]
3289 By default it checks for corruption in the low
3290 64k, making this memory unavailable for normal
3291 use. Use this parameter to scan for
3292 corruption in more or less memory.
3294 memory_corruption_check_period=seconds [X86,EARLY]
3295 By default it checks for corruption every 60
3296 seconds. Use this parameter to check at some
3297 other rate. 0 disables periodic checking.
3299 memory_hotplug.memmap_on_memory
3300 [KNL,X86,ARM] Boolean flag to enable this feature.
3301 Format: {on | off (default)}
3302 When enabled, runtime hotplugged memory will
3303 allocate its internal metadata (struct pages,
3304 those vmemmap pages cannot be optimized even
3305 if hugetlb_free_vmemmap is enabled) from the
3306 hotadded memory which will allow to hotadd a
3307 lot of memory without requiring additional
3309 This feature is disabled by default because it
3310 has some implication on large (e.g. GB)
3311 allocations in some configurations (e.g. small
3313 The state of the flag can be read in
3314 /sys/module/memory_hotplug/parameters/memmap_on_memory.
3315 Note that even when enabled, there are a few cases where
3316 the feature is not effective.
3318 memtest= [KNL,X86,ARM,M68K,PPC,RISCV,EARLY] Enable memtest
3320 default : 0 <disable>
3321 Specifies the number of memtest passes to be
3322 performed. Each pass selects another test
3323 pattern from a given set of patterns. Memtest
3324 fills the memory with this pattern, validates
3325 memory contents and reserves bad memory
3326 regions that are detected.
3328 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
3329 Valid arguments: on, off
3330 Default (depends on kernel configuration option):
3331 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
3332 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
3333 mem_encrypt=on: Activate SME
3334 mem_encrypt=off: Do not activate SME
3336 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst
3337 for details on when memory encryption can be activated.
3339 mem_sleep_default= [SUSPEND] Default system suspend mode:
3340 s2idle - Suspend-To-Idle
3341 shallow - Power-On Suspend or equivalent (if supported)
3342 deep - Suspend-To-RAM or equivalent (if supported)
3343 See Documentation/admin-guide/pm/sleep-states.rst.
3345 mfgpt_irq= [IA-32] Specify the IRQ to use for the
3346 Multi-Function General Purpose Timers on AMD Geode
3349 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
3350 the BIOS has incorrectly applied a workaround. TinyBIOS
3351 version 0.98 is known to be affected, 0.99 fixes the
3352 problem by letting the user disable the workaround.
3356 microcode.force_minrev= [X86]
3358 Enable or disable the microcode minimal revision
3359 enforcement for the runtime microcode loader.
3361 min_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory below this
3362 physical address is ignored.
3364 mini2440= [ARM,HW,KNL]
3365 Format:[0..2][b][c][t]
3367 MINI2440 configuration specification:
3368 0 - The attached screen is the 3.5" TFT
3369 1 - The attached screen is the 7" TFT
3370 2 - The VGA Shield is attached (1024x768)
3371 Leaving out the screen size parameter will not load
3372 the TFT driver, and the framebuffer will be left
3374 b - Enable backlight. The TFT backlight pin will be
3375 linked to the kernel VESA blanking code and a GPIO
3376 LED. This parameter is not necessary when using the
3378 c - Enable the s3c camera interface.
3379 t - Reserved for enabling touchscreen support. The
3380 touchscreen support is not enabled in the mainstream
3381 kernel as of 2.6.30, a preliminary port can be found
3382 in the "bleeding edge" mini2440 support kernel at
3383 https://repo.or.cz/w/linux-2.6/mini2440.git
3386 [X86,PPC,S390,ARM64,EARLY] Control optional mitigations for
3387 CPU vulnerabilities. This is a set of curated,
3388 arch-independent options, each of which is an
3389 aggregation of existing arch-specific options.
3392 Disable all optional CPU mitigations. This
3393 improves system performance, but it may also
3394 expose users to several CPU vulnerabilities.
3395 Equivalent to: if nokaslr then kpti=0 [ARM64]
3396 gather_data_sampling=off [X86]
3397 kvm.nx_huge_pages=off [X86]
3400 mmio_stale_data=off [X86]
3401 no_entry_flush [PPC]
3402 no_uaccess_flush [PPC]
3405 nospectre_bhb [ARM64]
3406 nospectre_v1 [X86,PPC]
3407 nospectre_v2 [X86,PPC,S390,ARM64]
3409 spec_store_bypass_disable=off [X86,PPC]
3410 spectre_v2_user=off [X86]
3411 srbds=off [X86,INTEL]
3412 ssbd=force-off [ARM64]
3413 tsx_async_abort=off [X86]
3416 This does not have any effect on
3417 kvm.nx_huge_pages when
3418 kvm.nx_huge_pages=force.
3421 Mitigate all CPU vulnerabilities, but leave SMT
3422 enabled, even if it's vulnerable. This is for
3423 users who don't want to be surprised by SMT
3424 getting disabled across kernel upgrades, or who
3425 have other ways of avoiding SMT-based attacks.
3426 Equivalent to: (default behavior)
3429 Mitigate all CPU vulnerabilities, disabling SMT
3430 if needed. This is for users who always want to
3431 be fully mitigated, even if it means losing SMT.
3432 Equivalent to: l1tf=flush,nosmt [X86]
3433 mds=full,nosmt [X86]
3434 tsx_async_abort=full,nosmt [X86]
3435 mmio_stale_data=full,nosmt [X86]
3436 retbleed=auto,nosmt [X86]
3439 [KNL,EARLY] When CONFIG_DEBUG_MEMORY_INIT is set, this
3440 parameter allows control of the logging verbosity for
3441 the additional memory initialisation checks. A value
3442 of 0 disables mminit logging and a level of 4 will
3443 log everything. Information is printed at KERN_DEBUG
3444 so loglevel=8 may also need to be specified.
3447 [X86,INTEL,EARLY] Control mitigation for the Processor
3448 MMIO Stale Data vulnerabilities.
3450 Processor MMIO Stale Data is a class of
3451 vulnerabilities that may expose data after an MMIO
3452 operation. Exposed data could originate or end in
3453 the same CPU buffers as affected by MDS and TAA.
3454 Therefore, similar to MDS and TAA, the mitigation
3455 is to clear the affected CPU buffers.
3457 This parameter controls the mitigation. The
3460 full - Enable mitigation on vulnerable CPUs
3462 full,nosmt - Enable mitigation and disable SMT on
3465 off - Unconditionally disable mitigation
3467 On MDS or TAA affected machines,
3468 mmio_stale_data=off can be prevented by an active
3469 MDS or TAA mitigation as these vulnerabilities are
3470 mitigated with the same mechanism so in order to
3471 disable this mitigation, you need to specify
3472 mds=off and tsx_async_abort=off too.
3474 Not specifying this option is equivalent to
3475 mmio_stale_data=full.
3478 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst
3480 <module>.async_probe[=<bool>] [KNL]
3481 If no <bool> value is specified or if the value
3482 specified is not a valid <bool>, enable asynchronous
3483 probe on this module. Otherwise, enable/disable
3484 asynchronous probe on this module as indicated by the
3485 <bool> value. See also: module.async_probe
3487 module.async_probe=<bool>
3488 [KNL] When set to true, modules will use async probing
3489 by default. To enable/disable async probing for a
3490 specific module, use the module specific control that
3491 is documented under <module>.async_probe. When both
3492 module.async_probe and <module>.async_probe are
3493 specified, <module>.async_probe takes precedence for
3494 the specific module.
3496 module.enable_dups_trace
3497 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set,
3498 this means that duplicate request_module() calls will
3499 trigger a WARN_ON() instead of a pr_warn(). Note that
3500 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s
3501 will always be issued and this option does nothing.
3503 [KNL] When CONFIG_MODULE_SIG is set, this means that
3504 modules without (valid) signatures will fail to load.
3505 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3506 is always true, so this option does nothing.
3508 module_blacklist= [KNL] Do not load a comma-separated list of
3509 modules. Useful for debugging problem modules.
3512 [MOUSE] Maximum time between finger touching and
3513 leaving touchpad surface for touch to be considered
3514 a tap and be reported as a left button click (for
3515 touchpads working in absolute mode only).
3517 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3518 reporting absolute coordinates, such as tablets
3519 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3520 reporting absolute coordinates, such as tablets
3522 movablecore= [KNL,X86,IA-64,PPC,EARLY]
3523 Format: nn[KMGTPE] | nn%
3524 This parameter is the complement to kernelcore=, it
3525 specifies the amount of memory used for migratable
3526 allocations. If both kernelcore and movablecore is
3527 specified, then kernelcore will be at *least* the
3528 specified value but may be more. If movablecore on its
3529 own is specified, the administrator must be careful
3530 that the amount of memory usable for all allocations
3533 movable_node [KNL,EARLY] Boot-time switch to make hotplugable memory
3534 NUMA nodes to be movable. This means that the memory
3535 of such nodes will be usable only for movable
3536 allocations which rules out almost all kernel
3537 allocations. Use with caution!
3539 MTD_Partition= [MTD]
3540 Format: <name>,<region-number>,<size>,<offset>
3542 MTD_Region= [MTD] Format:
3543 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3546 See drivers/mtd/parsers/cmdlinepart.c
3549 ARM/S3C2412 JIVE boot control
3551 See arch/arm/mach-s3c/mach-jive.c
3553 mtouchusb.raw_coordinates=
3554 [HW] Make the MicroTouch USB driver use raw coordinates
3555 ('y', default) or cooked coordinates ('n')
3557 mtrr=debug [X86,EARLY]
3558 Enable printing debug information related to MTRR
3559 registers at boot time.
3561 mtrr_chunk_size=nn[KMG,X86,EARLY]
3562 used for mtrr cleanup. It is largest continuous chunk
3563 that could hold holes aka. UC entries.
3565 mtrr_gran_size=nn[KMG,X86,EARLY]
3566 Used for mtrr cleanup. It is granularity of mtrr block.
3568 Large value could prevent small alignment from
3571 mtrr_spare_reg_nr=n [X86,EARLY]
3573 Range: 0,7 : spare reg number
3575 Used for mtrr cleanup. It is spare mtrr entries number.
3576 Set to 2 or more if your graphical card needs more.
3578 multitce=off [PPC] This parameter disables the use of the pSeries
3579 firmware feature for updating multiple TCE entries
3582 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3584 netdev= [NET] Network devices parameters
3585 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3586 Note that mem_start is often overloaded to mean
3587 something different and driver-specific.
3588 This usage is only documented in each driver source
3591 netpoll.carrier_timeout=
3592 [NET] Specifies amount of time (in seconds) that
3593 netpoll should wait for a carrier. By default netpoll
3597 [NETFILTER] Enable connection tracking flow accounting
3598 0 to disable accounting
3599 1 to enable accounting
3603 [NFS] sets the pathname to the program which is used
3604 to update the NFS client cache entries.
3606 nfs.cache_getent_timeout=
3607 [NFS] sets the timeout after which an attempt to
3608 update a cache entry is deemed to have failed.
3610 nfs.callback_nr_threads=
3611 [NFSv4] set the total number of threads that the
3612 NFS client will assign to service NFSv4 callback
3615 nfs.callback_tcpport=
3616 [NFS] set the TCP port on which the NFSv4 callback
3617 channel should listen.
3620 [NFS] specifies the number of times the NFSv4 client
3621 retries the request before returning an EAGAIN error,
3622 after a reply of NFS4ERR_DELAY from the server.
3623 Only applies if the softerr mount option is enabled,
3624 and the specified value is >= 0.
3627 [NFS] enable 64-bit inode numbers.
3628 If zero, the NFS client will fake up a 32-bit inode
3629 number for the readdir() and stat() syscalls instead
3630 of returning the full 64-bit number.
3631 The default is to return 64-bit inode numbers.
3633 nfs.idmap_cache_timeout=
3634 [NFS] set the maximum lifetime for idmapper cache
3637 nfs.max_session_cb_slots=
3638 [NFSv4.1] Sets the maximum number of session
3639 slots the client will assign to the callback
3640 channel. This determines the maximum number of
3641 callbacks the client will process in parallel for
3642 a particular server.
3644 nfs.max_session_slots=
3645 [NFSv4.1] Sets the maximum number of session slots
3646 the client will attempt to negotiate with the server.
3647 This limits the number of simultaneous RPC requests
3648 that the client can send to the NFSv4.1 server.
3649 Note that there is little point in setting this
3650 value higher than the max_tcp_slot_table_limit.
3652 nfs.nfs4_disable_idmapping=
3653 [NFSv4] When set to the default of '1', this option
3654 ensures that both the RPC level authentication
3655 scheme and the NFS level operations agree to use
3656 numeric uids/gids if the mount is using the
3657 'sec=sys' security flavour. In effect it is
3658 disabling idmapping, which can make migration from
3659 legacy NFSv2/v3 systems to NFSv4 easier.
3660 Servers that do not support this mode of operation
3661 will be autodetected by the client, and it will fall
3662 back to using the idmapper.
3663 To turn off this behaviour, set the value to '0'.
3666 [NFS4] Specify an additional fixed unique ident-
3667 ification string that NFSv4 clients can insert into
3668 their nfs_client_id4 string. This is typically a
3669 UUID that is generated at system install time.
3671 nfs.recover_lost_locks=
3672 [NFSv4] Attempt to recover locks that were lost due
3673 to a lease timeout on the server. Please note that
3674 doing this risks data corruption, since there are
3675 no guarantees that the file will remain unchanged
3676 after the locks are lost.
3677 If you want to enable the kernel legacy behaviour of
3678 attempting to recover these locks, then set this
3680 The default parameter value of '0' causes the kernel
3681 not to attempt recovery of lost locks.
3683 nfs.send_implementation_id=
3684 [NFSv4.1] Send client implementation identification
3685 information in exchange_id requests.
3686 If zero, no implementation identification information
3688 The default is to send the implementation identification
3691 nfs4.layoutstats_timer=
3692 [NFSv4.2] Change the rate at which the kernel sends
3693 layoutstats to the pNFS metadata server.
3695 Setting this to value to 0 causes the kernel to use
3696 whatever value is the default set by the layout
3697 driver. A non-zero value sets the minimum interval
3698 in seconds between layoutstats transmissions.
3700 nfsd.inter_copy_offload_enable=
3701 [NFSv4.2] When set to 1, the server will support
3702 server-to-server copies for which this server is
3703 the destination of the copy.
3705 nfsd.nfs4_disable_idmapping=
3706 [NFSv4] When set to the default of '1', the NFSv4
3707 server will return only numeric uids and gids to
3708 clients using auth_sys, and will accept numeric uids
3709 and gids from such clients. This is intended to ease
3710 migration from NFSv2/v3.
3712 nfsd.nfsd4_ssc_umount_timeout=
3713 [NFSv4.2] When used as the destination of a
3714 server-to-server copy, knfsd temporarily mounts
3715 the source server. It caches the mount in case
3716 it will be needed again, and discards it if not
3717 used for the number of milliseconds specified by
3720 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3721 See Documentation/admin-guide/nfs/nfsroot.rst.
3723 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3724 See Documentation/admin-guide/nfs/nfsroot.rst.
3726 nfsrootdebug [NFS] enable nfsroot debugging messages.
3727 See Documentation/admin-guide/nfs/nfsroot.rst.
3729 nmi_backtrace.backtrace_idle [KNL]
3730 Dump stacks even of idle CPUs in response to an
3731 NMI stack-backtrace request.
3733 nmi_debug= [KNL,SH] Specify one or more actions to take
3734 when a NMI is triggered.
3735 Format: [state][,regs][,debounce][,die]
3737 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3738 Format: [panic,][nopanic,][num]
3740 0 - turn hardlockup detector in nmi_watchdog off
3741 1 - turn hardlockup detector in nmi_watchdog on
3742 When panic is specified, panic when an NMI watchdog
3743 timeout occurs (or 'nopanic' to not panic on an NMI
3744 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3745 To disable both hard and soft lockup detectors,
3746 please see 'nowatchdog'.
3747 This is useful when you use a panic=... timeout and
3748 need the box quickly up again.
3750 These settings can be accessed at runtime via
3751 the nmi_watchdog and hardlockup_panic sysctls.
3753 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3754 emulation library even if a 387 maths coprocessor
3757 no4lvl [RISCV,EARLY] Disable 4-level and 5-level paging modes.
3758 Forces kernel to use 3-level paging instead.
3760 no5lvl [X86-64,RISCV,EARLY] Disable 5-level paging mode. Forces
3761 kernel to use 4-level paging instead.
3763 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3764 caches in the slab allocator. Saves per-node memory,
3765 but will impact performance.
3769 noaltinstr [S390,EARLY] Disables alternative instructions
3770 patching (CPU alternatives feature).
3772 noapic [SMP,APIC,EARLY] Tells the kernel to not make use of any
3773 IOAPICs that may be present in the system.
3775 noautogroup Disable scheduler automatic task group creation.
3780 [HW] Never suspend the console
3781 Disable suspending of consoles during suspend and
3782 hibernate operations. Once disabled, debugging
3783 messages can reach various consoles while the rest
3784 of the system is being put to sleep (ie, while
3785 debugging driver suspend/resume hooks). This may
3786 not work reliably with all consoles, but is known
3787 to work with serial and VGA consoles.
3788 To facilitate more flexible debugging, we also add
3789 console_suspend, a printk module parameter to control
3790 it. Users could use console_suspend (usually
3791 /sys/module/printk/parameters/console_suspend) to
3792 turn on/off it dynamically.
3795 [KNL,EARLY] Disable object debugging
3797 nodsp [SH] Disable hardware DSP at boot time.
3799 noefi [EFI,EARLY] Disable EFI runtime services support.
3801 no_entry_flush [PPC,EARLY] Don't flush the L1-D cache when entering the kernel.
3806 This affects only 32-bit executables.
3807 noexec32=on: enable non-executable mappings (default)
3808 read doesn't imply executable mappings
3809 noexec32=off: disable non-executable mappings
3810 read implies executable mappings
3812 no_file_caps Tells the kernel not to honor file capabilities. The
3813 only way then for a file to be executed with privilege
3814 is to be setuid root or executed by root.
3816 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3818 nofsgsbase [X86] Disables FSGSBASE instructions.
3820 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3821 register save and restore. The kernel will only save
3822 legacy floating-point registers on task switch.
3824 nohalt [IA-64] Tells the kernel not to use the power saving
3825 function PAL_HALT_LIGHT when idle. This increases
3826 power-consumption. On the positive side, it reduces
3827 interrupt wake-up latency, which may improve performance
3828 in certain environments such as networked servers or
3833 Force pointers printed to the console or buffers to be
3834 unhashed. By default, when a pointer is printed via %p
3835 format string, that pointer is "hashed", i.e. obscured
3836 by hashing the pointer value. This is a security feature
3837 that hides actual kernel addresses from unprivileged
3838 users, but it also makes debugging the kernel more
3839 difficult since unequal pointers can no longer be
3840 compared. However, if this command-line option is
3841 specified, then all normal pointers will have their true
3842 value printed. This option should only be specified when
3843 debugging the kernel. Please do not use on production
3846 nohibernate [HIBERNATION] Disable hibernation and resume.
3848 nohlt [ARM,ARM64,MICROBLAZE,MIPS,PPC,SH] Forces the kernel to
3849 busy wait in do_idle() and not use the arch_cpu_idle()
3850 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3851 to be effective. This is useful on platforms where the
3852 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3853 correctly or when doing power measurements to evaluate
3854 the impact of the sleep instructions. This is also
3855 useful when using JTAG debugger.
3857 nohugeiomap [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge I/O mappings.
3859 nohugevmalloc [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge vmalloc mappings.
3861 nohz= [KNL] Boottime enable/disable dynamic ticks
3862 Valid arguments: on, off
3865 nohz_full= [KNL,BOOT,SMP,ISOL]
3866 The argument is a cpu list, as described above.
3867 In kernels built with CONFIG_NO_HZ_FULL=y, set
3868 the specified list of CPUs whose tick will be stopped
3869 whenever possible. The boot CPU will be forced outside
3870 the range to maintain the timekeeping. Any CPUs
3871 in this list will have their RCU callbacks offloaded,
3872 just as if they had also been called out in the
3873 rcu_nocbs= boot parameter.
3875 Note that this argument takes precedence over
3876 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
3878 noinitrd [RAM] Tells the kernel not to load any configured
3881 nointremap [X86-64,Intel-IOMMU,EARLY] Do not enable interrupt
3883 [Deprecated - use intremap=off]
3887 noinvpcid [X86,EARLY] Disable the INVPCID cpu feature.
3889 noiotrap [SH] Disables trapped I/O port accesses.
3891 noirqdebug [X86-32] Disables the code which attempts to detect and
3892 disable unhandled interrupt sources.
3894 noisapnp [ISAPNP] Disables ISA PnP code.
3896 nojitter [IA-64] Disables jitter checking for ITC timers.
3899 When CONFIG_RANDOMIZE_BASE is set, this disables
3900 kernel and module base offset ASLR (Address Space
3901 Layout Randomization).
3903 no-kvmapf [X86,KVM,EARLY] Disable paravirtualized asynchronous page
3906 no-kvmclock [X86,KVM,EARLY] Disable paravirtualized KVM clock driver
3908 nolapic [X86-32,APIC,EARLY] Do not enable or use the local APIC.
3910 nolapic_timer [X86-32,APIC,EARLY] Do not use the local APIC timer.
3912 nomca [IA-64] Disable machine check abort handling
3914 nomce [X86-32] Disable Machine Check Exception
3916 nomfgpt [X86-32] Disable Multi-Function General Purpose
3917 Timer usage (for AMD Geode machines).
3919 nomodeset Disable kernel modesetting. Most systems' firmware
3920 sets up a display mode and provides framebuffer memory
3921 for output. With nomodeset, DRM and fbdev drivers will
3922 not load if they could possibly displace the pre-
3923 initialized output. Only the system framebuffer will
3924 be available for use. The respective drivers will not
3925 perform display-mode changes or accelerated rendering.
3927 Useful as error fallback, or for testing and debugging.
3929 nomodule Disable module load
3931 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3932 shutdown the other cpus. Instead use the REBOOT_VECTOR
3935 nopat [X86,EARLY] Disable PAT (page attribute table extension of
3936 pagetables) support.
3938 nopcid [X86-64,EARLY] Disable the PCID cpu feature.
3940 nopku [X86] Disable Memory Protection Keys CPU feature found
3943 nopti [X86-64,EARLY]
3944 Equivalent to pti=off
3946 nopv= [X86,XEN,KVM,HYPER_V,VMWARE,EARLY]
3947 Disables the PV optimizations forcing the guest to run
3948 as generic guest with no PV drivers. Currently support
3949 XEN HVM, KVM, HYPER_V and VMWARE guest.
3951 nopvspin [X86,XEN,KVM,EARLY]
3952 Disables the qspinlock slow path using PV optimizations
3953 which allow the hypervisor to 'idle' the guest on lock
3956 norandmaps Don't use address space randomization. Equivalent to
3957 echo 0 > /proc/sys/kernel/randomize_va_space
3959 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3960 with UP alternatives
3962 noresume [SWSUSP] Disables resume and restores original swap
3967 no-scroll [VGA] Disables scrollback.
3968 This is required for the Braillex ib80-piezo Braille
3969 reader made by F.H. Papenmeier (Germany).
3971 nosgx [X86-64,SGX,EARLY] Disables Intel SGX kernel support.
3974 Disable SMAP (Supervisor Mode Access Prevention)
3975 even if it is supported by processor.
3977 nosmep [PPC64s,EARLY]
3978 Disable SMEP (Supervisor Mode Execution Prevention)
3979 even if it is supported by processor.
3981 nosmp [SMP,EARLY] Tells an SMP kernel to act as a UP kernel,
3982 and disable the IO APIC. legacy for "maxcpus=0".
3984 nosmt [KNL,MIPS,PPC,S390,EARLY] Disable symmetric multithreading (SMT).
3985 Equivalent to smt=1.
3987 [KNL,X86,PPC] Disable symmetric multithreading (SMT).
3988 nosmt=force: Force disable SMT, cannot be undone
3989 via the sysfs control file.
3991 nosoftlockup [KNL] Disable the soft-lockup detector.
3993 nospec_store_bypass_disable
3994 [HW,EARLY] Disable all mitigations for the Speculative
3995 Store Bypass vulnerability
3997 nospectre_bhb [ARM64,EARLY] Disable all mitigations for Spectre-BHB (branch
3998 history injection) vulnerability. System may allow data leaks
4001 nospectre_v1 [X86,PPC,EARLY] Disable mitigations for Spectre Variant 1
4002 (bounds check bypass). With this option data leaks are
4003 possible in the system.
4005 nospectre_v2 [X86,PPC_E500,ARM64,EARLY] Disable all mitigations
4006 for the Spectre variant 2 (indirect branch
4007 prediction) vulnerability. System may allow data
4008 leaks with this option.
4010 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES,RISCV,EARLY] Disable
4011 paravirtualized steal time accounting. steal time is
4012 computed, but won't influence scheduler behaviour
4014 nosync [HW,M68K] Disables sync negotiation for all devices.
4016 no_timer_check [X86,APIC] Disables the code which tests for
4017 broken timer IRQ sources.
4020 [PPC,EARLY] Don't flush the L1-D cache after accessing user data.
4022 novmcoredd [KNL,KDUMP]
4023 Disable device dump. Device dump allows drivers to
4024 append dump data to vmcore so you can collect driver
4025 specified debug info. Drivers can append the data
4026 without any limit and this data is stored in memory,
4027 so this may cause significant memory stress. Disabling
4028 device dump can help save memory but the driver debug
4029 data will be no longer available. This parameter
4030 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
4034 [X86,PV_OPS,EARLY] Disable paravirtualized VMware
4035 scheduler clock and use the default one.
4037 nowatchdog [KNL] Disable both lockup detectors, i.e.
4038 soft-lockup and NMI watchdog (hard-lockup).
4042 nox2apic [X86-64,APIC,EARLY] Do not enable x2APIC mode.
4044 NOTE: this parameter will be ignored on systems with the
4045 LEGACY_XAPIC_DISABLED bit set in the
4046 IA32_XAPIC_DISABLE_STATUS MSR.
4048 noxsave [BUGS=X86] Disables x86 extended register state save
4049 and restore using xsave. The kernel will fallback to
4050 enabling legacy floating-point and sse state.
4052 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
4053 register states. The kernel will fall back to use
4054 xsave to save the states. By using this parameter,
4055 performance of saving the states is degraded because
4056 xsave doesn't support modified optimization while
4057 xsaveopt supports it on xsaveopt enabled systems.
4059 noxsaves [X86] Disables xsaves and xrstors used in saving and
4060 restoring x86 extended register state in compacted
4061 form of xsave area. The kernel will fall back to use
4062 xsaveopt and xrstor to save and restore the states
4063 in standard form of xsave area. By using this
4064 parameter, xsave area per process might occupy more
4065 memory on xsaves enabled systems.
4067 nps_mtm_hs_ctr= [KNL,ARC]
4068 This parameter sets the maximum duration, in
4069 cycles, each HW thread of the CTOP can run
4070 without interruptions, before HW switches it.
4071 The actual maximum duration is 16 times this
4073 Format: integer between 1 and 255
4076 nptcg= [IA-64] Override max number of concurrent global TLB
4077 purges which is reported from either PAL_VM_SUMMARY or
4080 nr_cpus= [SMP,EARLY] Maximum number of processors that an SMP kernel
4081 could support. nr_cpus=n : n >= 1 limits the kernel to
4082 support 'n' processors. It could be larger than the
4083 number of already plugged CPU during bootup, later in
4084 runtime you can physically add extra cpu until it reaches
4085 n. So during boot up some boot time memory for per-cpu
4086 variables need be pre-allocated for later physical cpu
4089 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
4091 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86, EARLY]
4092 Disable NUMA, Only set up a single NUMA node
4093 spanning all memory.
4095 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
4097 Allowed values are enable and disable
4099 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
4100 'node', 'default' can be specified
4101 This can be set from sysctl after boot.
4102 See Documentation/admin-guide/sysctl/vm.rst for details.
4104 ohci1394_dma=early [HW,EARLY] enable debugging via the ohci1394 driver.
4105 See Documentation/core-api/debugging-via-ohci1394.rst for more
4108 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
4109 Rather than timing out after 20 ms if an EC
4110 command is not properly ACKed, override the length
4111 of the timeout. We have interrupts disabled while
4112 waiting for the ACK, so if this is set too high
4113 interrupts *may* be lost!
4115 omap_mux= [OMAP] Override bootloader pin multiplexing.
4116 Format: <mux_mode0.mode_name=value>...
4117 For example, to override I2C bus2:
4118 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
4120 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
4122 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
4124 boundary - index of last SLC block on Flex-OneNAND.
4125 The remaining blocks are configured as MLC blocks.
4126 lock - Configure if Flex-OneNAND boundary should be locked.
4127 Once locked, the boundary cannot be changed.
4128 1 indicates lock status, 0 indicates unlock status.
4130 oops=panic [KNL,EARLY]
4131 Always panic on oopses. Default is to just kill the
4132 process, but there is a small probability of
4133 deadlocking the machine.
4134 This will also cause panics on machine check exceptions.
4135 Useful together with panic=30 to trigger a reboot.
4138 [KNL] Boolean flag to control whether the page allocator
4139 should randomize its free lists. The randomization may
4140 be automatically enabled if the kernel detects it is
4141 running on a platform with a direct-mapped memory-side
4142 cache, and this parameter can be used to
4143 override/disable that behavior. The state of the flag
4144 can be read from sysfs at:
4145 /sys/module/page_alloc/parameters/shuffle.
4147 page_owner= [KNL,EARLY] Boot-time page_owner enabling option.
4148 Storage of the information about who allocated
4149 each page is disabled in default. With this switch,
4151 on: enable the feature
4153 page_poison= [KNL,EARLY] Boot-time parameter changing the state of
4154 poisoning on the buddy allocator, available with
4155 CONFIG_PAGE_POISONING=y.
4156 off: turn off poisoning (default)
4157 on: turn on poisoning
4159 page_reporting.page_reporting_order=
4160 [KNL] Minimal page reporting order
4162 Adjust the minimal page reporting order. The page
4163 reporting is disabled when it exceeds MAX_PAGE_ORDER.
4165 panic= [KNL] Kernel behaviour on panic: delay <timeout>
4166 timeout > 0: seconds before rebooting
4167 timeout = 0: wait forever
4168 timeout < 0: reboot immediately
4171 panic_on_taint= [KNL,EARLY]
4172 Bitmask for conditionally calling panic() in add_taint()
4173 Format: <hex>[,nousertaint]
4174 Hexadecimal bitmask representing the set of TAINT flags
4175 that will cause the kernel to panic when add_taint() is
4176 called with any of the flags in this set.
4177 The optional switch "nousertaint" can be utilized to
4178 prevent userspace forced crashes by writing to sysctl
4179 /proc/sys/kernel/tainted any flagset matching with the
4180 bitmask set on panic_on_taint.
4181 See Documentation/admin-guide/tainted-kernels.rst for
4182 extra details on the taint flags that users can pick
4183 to compose the bitmask to assign to panic_on_taint.
4185 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump
4188 panic_print= Bitmask for printing system info when panic happens.
4189 User can chose combination of the following bits:
4190 bit 0: print all tasks info
4191 bit 1: print system memory info
4192 bit 2: print timer info
4193 bit 3: print locks info if CONFIG_LOCKDEP is on
4194 bit 4: print ftrace buffer
4195 bit 5: print all printk messages in buffer
4196 bit 6: print all CPUs backtrace (if available in the arch)
4197 *Be aware* that this option may print a _lot_ of lines,
4198 so there are risks of losing older messages in the log.
4199 Use this option carefully, maybe worth to setup a
4200 bigger log buffer with "log_buf_len" along with this.
4202 parkbd.port= [HW] Parallel port number the keyboard adapter is
4203 connected to, default is 0.
4205 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
4206 0 for XT, 1 for AT (default is AT).
4209 parport= [HW,PPT] Specify parallel ports. 0 disables.
4210 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
4211 Use 'auto' to force the driver to use any
4212 IRQ/DMA settings detected (the default is to
4213 ignore detected IRQ/DMA settings because of
4214 possible conflicts). You can specify the base
4215 address, IRQ, and DMA settings; IRQ and DMA
4216 should be numbers, or 'auto' (for using detected
4217 settings on that particular port), or 'nofifo'
4218 (to avoid using a FIFO even if it is detected).
4219 Parallel ports are assigned in the order they
4220 are specified on the command line, starting
4223 parport_init_mode= [HW,PPT]
4224 Configure VIA parallel port to operate in
4225 a specific mode. This is necessary on Pegasos
4226 computer where firmware has no options for setting
4227 up parallel port mode and sets it to spp.
4228 Currently this function knows 686a and 8231 chips.
4229 Format: [spp|ps2|epp|ecp|ecpepp]
4231 pata_legacy.all= [HW,LIBATA]
4233 Set to non-zero to probe primary and secondary ISA
4234 port ranges on PCI systems where no PCI PATA device
4235 has been found at either range. Disabled by default.
4237 pata_legacy.autospeed= [HW,LIBATA]
4239 Set to non-zero if a chip is present that snoops speed
4240 changes. Disabled by default.
4242 pata_legacy.ht6560a= [HW,LIBATA]
4244 Set to 1, 2, or 3 for HT 6560A on the primary channel,
4245 the secondary channel, or both channels respectively.
4246 Disabled by default.
4248 pata_legacy.ht6560b= [HW,LIBATA]
4250 Set to 1, 2, or 3 for HT 6560B on the primary channel,
4251 the secondary channel, or both channels respectively.
4252 Disabled by default.
4254 pata_legacy.iordy_mask= [HW,LIBATA]
4256 IORDY enable mask. Set individual bits to allow IORDY
4257 for the respective channel. Bit 0 is for the first
4258 legacy channel handled by this driver, bit 1 is for
4259 the second channel, and so on. The sequence will often
4260 correspond to the primary legacy channel, the secondary
4261 legacy channel, and so on, but the handling of a PCI
4262 bus and the use of other driver options may interfere
4263 with the sequence. By default IORDY is allowed across
4266 pata_legacy.opti82c46x= [HW,LIBATA]
4268 Set to 1, 2, or 3 for Opti 82c611A on the primary
4269 channel, the secondary channel, or both channels
4270 respectively. Disabled by default.
4272 pata_legacy.opti82c611a= [HW,LIBATA]
4274 Set to 1, 2, or 3 for Opti 82c465MV on the primary
4275 channel, the secondary channel, or both channels
4276 respectively. Disabled by default.
4278 pata_legacy.pio_mask= [HW,LIBATA]
4280 PIO mode mask for autospeed devices. Set individual
4281 bits to allow the use of the respective PIO modes.
4282 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
4283 All modes allowed by default.
4285 pata_legacy.probe_all= [HW,LIBATA]
4287 Set to non-zero to probe tertiary and further ISA
4288 port ranges on PCI systems. Disabled by default.
4290 pata_legacy.probe_mask= [HW,LIBATA]
4292 Probe mask for legacy ISA PATA ports. Depending on
4293 platform configuration and the use of other driver
4294 options up to 6 legacy ports are supported: 0x1f0,
4295 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
4296 of individual ports can be disabled by setting the
4297 corresponding bits in the mask to 1. Bit 0 is for
4298 the first port in the list above (0x1f0), and so on.
4299 By default all supported ports are probed.
4301 pata_legacy.qdi= [HW,LIBATA]
4303 Set to non-zero to probe QDI controllers. By default
4304 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
4306 pata_legacy.winbond= [HW,LIBATA]
4308 Set to non-zero to probe Winbond controllers. Use
4309 the standard I/O port (0x130) if 1, otherwise the
4310 value given is the I/O port to use (typically 0x1b0).
4311 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
4314 pata_platform.pio_mask= [HW,LIBATA]
4316 Supported PIO mode mask. Set individual bits to allow
4317 the use of the respective PIO modes. Bit 0 is for
4318 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
4322 Halt all CPUs after the first oops has been printed for
4323 the specified number of seconds. This is to be used if
4324 your oopses keep scrolling off the screen.
4328 pci=option[,option...] [PCI,EARLY] various PCI subsystem options.
4330 Some options herein operate on a specific device
4331 or a set of devices (<pci_dev>). These are
4332 specified in one of the following formats:
4334 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
4335 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
4337 Note: the first format specifies a PCI
4338 bus/device/function address which may change
4339 if new hardware is inserted, if motherboard
4340 firmware changes, or due to changes caused
4341 by other kernel parameters. If the
4342 domain is left unspecified, it is
4343 taken to be zero. Optionally, a path
4344 to a device through multiple device/function
4345 addresses can be specified after the base
4346 address (this is more robust against
4347 renumbering issues). The second format
4348 selects devices using IDs from the
4349 configuration space which may match multiple
4350 devices in the system.
4352 earlydump dump PCI config space before the kernel
4354 off [X86] don't probe for the PCI bus
4355 bios [X86-32] force use of PCI BIOS, don't access
4356 the hardware directly. Use this if your machine
4357 has a non-standard PCI host bridge.
4358 nobios [X86-32] disallow use of PCI BIOS, only direct
4359 hardware access methods are allowed. Use this
4360 if you experience crashes upon bootup and you
4361 suspect they are caused by the BIOS.
4362 conf1 [X86] Force use of PCI Configuration Access
4363 Mechanism 1 (config address in IO port 0xCF8,
4364 data in IO port 0xCFC, both 32-bit).
4365 conf2 [X86] Force use of PCI Configuration Access
4366 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
4367 the function, IO port 0xCFA, also 8-bit, sets
4368 bus number. The config space is then accessed
4369 through ports 0xC000-0xCFFF).
4370 See http://wiki.osdev.org/PCI for more info
4371 on the configuration access mechanisms.
4372 noaer [PCIE] If the PCIEAER kernel config parameter is
4373 enabled, this kernel boot option can be used to
4374 disable the use of PCIE advanced error reporting.
4375 nodomains [PCI] Disable support for multiple PCI
4376 root domains (aka PCI segments, in ACPI-speak).
4377 nommconf [X86] Disable use of MMCONFIG for PCI
4379 check_enable_amd_mmconf [X86] check for and enable
4380 properly configured MMIO access to PCI
4381 config space on AMD family 10h CPU
4382 nomsi [MSI] If the PCI_MSI kernel config parameter is
4383 enabled, this kernel boot option can be used to
4384 disable the use of MSI interrupts system-wide.
4385 noioapicquirk [APIC] Disable all boot interrupt quirks.
4386 Safety option to keep boot IRQs enabled. This
4387 should never be necessary.
4388 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
4389 primary IO-APIC for bridges that cannot disable
4390 boot IRQs. This fixes a source of spurious IRQs
4391 when the system masks IRQs.
4392 noioapicreroute [APIC] Disable workaround that uses the
4393 boot IRQ equivalent of an IRQ that connects to
4394 a chipset where boot IRQs cannot be disabled.
4395 The opposite of ioapicreroute.
4396 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
4397 routing table. These calls are known to be buggy
4398 on several machines and they hang the machine
4399 when used, but on other computers it's the only
4400 way to get the interrupt routing table. Try
4401 this option if the kernel is unable to allocate
4402 IRQs or discover secondary PCI buses on your
4404 rom [X86] Assign address space to expansion ROMs.
4405 Use with caution as certain devices share
4406 address decoders between ROMs and other
4408 norom [X86] Do not assign address space to
4409 expansion ROMs that do not already have
4410 BIOS assigned address ranges.
4411 nobar [X86] Do not assign address space to the
4412 BARs that weren't assigned by the BIOS.
4413 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
4414 assigned automatically to PCI devices. You can
4415 make the kernel exclude IRQs of your ISA cards
4417 pirqaddr=0xAAAAA [X86] Specify the physical address
4418 of the PIRQ table (normally generated
4419 by the BIOS) if it is outside the
4420 F0000h-100000h range.
4421 lastbus=N [X86] Scan all buses thru bus #N. Can be
4422 useful if the kernel is unable to find your
4423 secondary buses and you want to tell it
4424 explicitly which ones they are.
4425 assign-busses [X86] Always assign all PCI bus
4426 numbers ourselves, overriding
4427 whatever the firmware may have done.
4428 usepirqmask [X86] Honor the possible IRQ mask stored
4429 in the BIOS $PIR table. This is needed on
4430 some systems with broken BIOSes, notably
4431 some HP Pavilion N5400 and Omnibook XE3
4432 notebooks. This will have no effect if ACPI
4433 IRQ routing is enabled.
4434 noacpi [X86] Do not use ACPI for IRQ routing
4435 or for PCI scanning.
4436 use_crs [X86] Use PCI host bridge window information
4437 from ACPI. On BIOSes from 2008 or later, this
4438 is enabled by default. If you need to use this,
4439 please report a bug.
4440 nocrs [X86] Ignore PCI host bridge windows from ACPI.
4441 If you need to use this, please report a bug.
4442 use_e820 [X86] Use E820 reservations to exclude parts of
4443 PCI host bridge windows. This is a workaround
4444 for BIOS defects in host bridge _CRS methods.
4445 If you need to use this, please report a bug to
4446 <linux-pci@vger.kernel.org>.
4447 no_e820 [X86] Ignore E820 reservations for PCI host
4448 bridge windows. This is the default on modern
4449 hardware. If you need to use this, please report
4450 a bug to <linux-pci@vger.kernel.org>.
4451 routeirq Do IRQ routing for all PCI devices.
4452 This is normally done in pci_enable_device(),
4453 so this option is a temporary workaround
4454 for broken drivers that don't call it.
4455 skip_isa_align [X86] do not align io start addr, so can
4456 handle more pci cards
4457 noearly [X86] Don't do any early type 1 scanning.
4458 This might help on some broken boards which
4459 machine check when some devices' config space
4460 is read. But various workarounds are disabled
4461 and some IOMMU drivers will not work.
4462 bfsort Sort PCI devices into breadth-first order.
4463 This sorting is done to get a device
4464 order compatible with older (<= 2.4) kernels.
4465 nobfsort Don't sort PCI devices into breadth-first order.
4466 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4467 tuning and use the BIOS-configured MPS defaults.
4468 pcie_bus_safe Set every device's MPS to the largest value
4469 supported by all devices below the root complex.
4470 pcie_bus_perf Set device MPS to the largest allowable MPS
4471 based on its parent bus. Also set MRRS (Max
4472 Read Request Size) to the largest supported
4473 value (no larger than the MPS that the device
4474 or bus can support) for best performance.
4475 pcie_bus_peer2peer Set every device's MPS to 128B, which
4476 every device is guaranteed to support. This
4477 configuration allows peer-to-peer DMA between
4478 any pair of devices, possibly at the cost of
4479 reduced performance. This also guarantees
4480 that hot-added devices will work.
4481 cbiosize=nn[KMG] The fixed amount of bus space which is
4482 reserved for the CardBus bridge's IO window.
4483 The default value is 256 bytes.
4484 cbmemsize=nn[KMG] The fixed amount of bus space which is
4485 reserved for the CardBus bridge's memory
4486 window. The default value is 64 megabytes.
4489 [<order of align>@]<pci_dev>[; ...]
4490 Specifies alignment and device to reassign
4491 aligned memory resources. How to
4492 specify the device is described above.
4493 If <order of align> is not specified,
4494 PAGE_SIZE is used as alignment.
4495 A PCI-PCI bridge can be specified if resource
4496 windows need to be expanded.
4497 To specify the alignment for several
4498 instances of a device, the PCI vendor,
4499 device, subvendor, and subdevice may be
4500 specified, e.g., 12@pci:8086:9c22:103c:198f
4501 for 4096-byte alignment.
4502 ecrc= Enable/disable PCIe ECRC (transaction layer
4503 end-to-end CRC checking). Only effective if
4504 OS has native AER control (either granted by
4505 ACPI _OSC or forced via "pcie_ports=native")
4506 bios: Use BIOS/firmware settings. This is the
4510 hpiosize=nn[KMG] The fixed amount of bus space which is
4511 reserved for hotplug bridge's IO window.
4512 Default size is 256 bytes.
4513 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4514 reserved for hotplug bridge's MMIO window.
4515 Default size is 2 megabytes.
4516 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4517 reserved for hotplug bridge's MMIO_PREF window.
4518 Default size is 2 megabytes.
4519 hpmemsize=nn[KMG] The fixed amount of bus space which is
4520 reserved for hotplug bridge's MMIO and
4522 Default size is 2 megabytes.
4523 hpbussize=nn The minimum amount of additional bus numbers
4524 reserved for buses below a hotplug bridge.
4526 realloc= Enable/disable reallocating PCI bridge resources
4527 if allocations done by BIOS are too small to
4528 accommodate resources required by all child
4530 off: Turn realloc off
4532 realloc same as realloc=on
4533 noari do not use PCIe ARI.
4534 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4535 do not use PCIe ATS (and IOMMU device IOTLB).
4536 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4537 only look for one device below a PCIe downstream
4539 big_root_window Try to add a big 64bit memory window to the PCIe
4540 root complex on AMD CPUs. Some GFX hardware
4541 can resize a BAR to allow access to all VRAM.
4542 Adding the window is slightly risky (it may
4543 conflict with unreported devices), so this
4545 disable_acs_redir=<pci_dev>[; ...]
4546 Specify one or more PCI devices (in the format
4547 specified above) separated by semicolons.
4548 Each device specified will have the PCI ACS
4549 redirect capabilities forced off which will
4550 allow P2P traffic between devices through
4551 bridges without forcing it upstream. Note:
4552 this removes isolation between devices and
4553 may put more devices in an IOMMU group.
4554 force_floating [S390] Force usage of floating interrupts.
4555 nomio [S390] Do not use MIO instructions.
4556 norid [S390] ignore the RID field and force use of
4557 one PCI domain per PCI function
4559 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4562 force Enable ASPM even on devices that claim not to support it.
4563 WARNING: Forcing ASPM on may cause system lockups.
4565 pcie_ports= [PCIE] PCIe port services handling:
4566 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4567 even if the platform doesn't give the OS permission to
4568 use them. This may cause conflicts if the platform
4569 also tries to use these services.
4570 dpc-native Use native PCIe service for DPC only. May
4571 cause conflicts if firmware uses AER or DPC.
4572 compat Disable native PCIe services (PME, AER, DPC, PCIe
4575 pcie_port_pm= [PCIE] PCIe port power management handling:
4576 off Disable power management of all PCIe ports
4577 force Forcibly enable power management of all PCIe ports
4579 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4580 nomsi Do not use MSI for native PCIe PME signaling (this makes
4581 all PCIe root ports use INTx for all services).
4583 pcmv= [HW,PCMCIA] BadgePAD 4
4587 Keep all power-domains already enabled by bootloader on,
4588 even if no driver has claimed them. This is useful
4589 for debug and development, but should not be
4590 needed on a platform with proper driver support.
4592 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4595 See arch/parisc/kernel/pdc_chassis.c
4597 percpu_alloc= [MM,EARLY]
4598 Select which percpu first chunk allocator to use.
4599 Currently supported values are "embed" and "page".
4600 Archs may support subset or none of the selections.
4601 See comments in mm/percpu.c for details on each
4602 allocator. This parameter is primarily for debugging
4603 and performance comparison.
4605 pirq= [SMP,APIC] Manual mp-table setup
4606 See Documentation/arch/x86/i386/IO-APIC.rst.
4608 plip= [PPT,NET] Parallel port network link
4609 Format: { parport<nr> | timid | 0 }
4610 See also Documentation/admin-guide/parport.rst.
4612 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4613 Override pmtimer IOPort with a hex value.
4616 pmu_override= [PPC] Override the PMU.
4617 This option takes over the PMU facility, so it is no
4618 longer usable by perf. Setting this option starts the
4619 PMU counters by setting MMCR0 to 0 (the FC bit is
4620 cleared). If a number is given, then MMCR1 is set to
4621 that number, otherwise (e.g., 'pmu_override=on'), MMCR1
4624 pm_debug_messages [SUSPEND,KNL]
4625 Enable suspend/resume debug messages during boot up.
4628 Enable PNP debug messages (depends on the
4629 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4630 via /sys/module/pnp/parameters/debug. We always show
4631 current resource usage; turning this on also shows
4632 possible settings and some assignment information.
4638 { on | off | curr | res | no-curr | no-res }
4641 [ISAPNP] Exclude IRQs for the autoconfiguration
4644 [ISAPNP] Exclude DMAs for the autoconfiguration
4646 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4647 Ranges are in pairs (I/O port base and size).
4650 [ISAPNP] Exclude memory regions for the
4652 Ranges are in pairs (memory base and size).
4654 ports= [IP_VS_FTP] IPVS ftp helper module
4656 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4658 Format: <port>,<port>....
4660 powersave=off [PPC] This option disables power saving features.
4661 It specifically disables cpuidle and sets the
4662 platform machine description specific power_save
4663 function to NULL. On Idle the CPU just reduces
4666 ppc_strict_facility_enable
4667 [PPC,ENABLE] This option catches any kernel floating point,
4668 Altivec, VSX and SPE outside of regions specifically
4669 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4670 There is some performance impact when enabling this.
4674 Disable Hardware Transactional Memory
4677 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4678 none - Limited to cond_resched() calls
4679 voluntary - Limited to cond_resched() and might_sleep() calls
4680 full - Any section that isn't explicitly preempt disabled
4681 can be preempted anytime.
4683 print-fatal-signals=
4684 [KNL] debug: print fatal signals
4686 If enabled, warn about various signal handling
4687 related application anomalies: too many signals,
4688 too many POSIX.1 timers, fatal signals causing a
4691 If you hit the warning due to signal overflow,
4692 you might want to try "ulimit -i unlimited".
4696 printk.always_kmsg_dump=
4697 Trigger kmsg_dump for cases other than kernel oops or
4699 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4702 printk.console_no_auto_verbose=
4703 Disable console loglevel raise on oops, panic
4704 or lockdep-detected issues (only if lock debug is on).
4705 With an exception to setups with low baudrate on
4706 serial console, keeping this 0 is a good choice
4707 in order to provide more debug information.
4709 default: 0 (auto_verbose is enabled)
4711 printk.devkmsg={on,off,ratelimit}
4712 Control writing to /dev/kmsg.
4713 on - unlimited logging to /dev/kmsg from userspace
4714 off - logging to /dev/kmsg disabled
4715 ratelimit - ratelimit the logging
4718 printk.time= Show timing data prefixed to each printk message line
4719 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4721 processor.max_cstate= [HW,ACPI]
4722 Limit processor to maximum C-state
4723 max_cstate=9 overrides any DMI blacklist limit.
4725 processor.nocst [HW,ACPI]
4726 Ignore the _CST method to determine C-states,
4727 instead using the legacy FADT method
4729 profile= [KNL] Enable kernel profiling via /proc/profile
4730 Format: [<profiletype>,]<number>
4731 Param: <profiletype>: "schedule", "sleep", or "kvm"
4732 [defaults to kernel profiling]
4733 Param: "schedule" - profile schedule points.
4734 Param: "sleep" - profile D-state sleeping (millisecs).
4735 Requires CONFIG_SCHEDSTATS
4736 Param: "kvm" - profile VM exits.
4737 Param: <number> - step/bucket size as a power of 2 for
4738 statistical time based profiling.
4740 prompt_ramdisk= [RAM] [Deprecated]
4742 prot_virt= [S390] enable hosting protected virtual machines
4743 isolated from the hypervisor (if hardware supports
4747 psi= [KNL] Enable or disable pressure stall information
4751 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4752 probe for; one of (bare|imps|exps|lifebook|any).
4753 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4755 psmouse.resetafter= [HW,MOUSE]
4756 Try to reset the device after so many bad packets
4759 [HW,MOUSE] Set desired mouse resolution, in dpi.
4760 psmouse.smartscroll=
4761 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4762 0 = disabled, 1 = enabled (default).
4764 pstore.backend= Specify the name of the pstore backend to use
4766 pti= [X86-64] Control Page Table Isolation of user and
4767 kernel address spaces. Disabling this feature
4768 removes hardening, but improves performance of
4769 system calls and interrupts.
4771 on - unconditionally enable
4772 off - unconditionally disable
4773 auto - kernel detects whether your CPU model is
4774 vulnerable to issues that PTI mitigates
4776 Not specifying this option is equivalent to pti=auto.
4779 [KNL] Number of legacy pty's. Overwrites compiled-in
4782 quiet [KNL,EARLY] Disable most log messages
4786 radix_hcall_invalidate=on [PPC/PSERIES]
4787 Disable RADIX GTSE feature and use hcall for TLB
4791 See Documentation/admin-guide/md.rst.
4793 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4794 See Documentation/admin-guide/blockdev/ramdisk.rst.
4796 ramdisk_start= [RAM] RAM disk image start address
4798 random.trust_cpu=off
4799 [KNL,EARLY] Disable trusting the use of the CPU's
4800 random number generator (if available) to
4801 initialize the kernel's RNG.
4803 random.trust_bootloader=off
4804 [KNL,EARLY] Disable trusting the use of the a seed
4805 passed by the bootloader (if available) to
4806 initialize the kernel's RNG.
4808 randomize_kstack_offset=
4809 [KNL,EARLY] Enable or disable kernel stack offset
4810 randomization, which provides roughly 5 bits of
4811 entropy, frustrating memory corruption attacks
4812 that depend on stack address determinism or
4813 cross-syscall address exposures. This is only
4814 available on architectures that have defined
4815 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4816 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4817 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4819 ras=option[,option,...] [KNL] RAS-specific options
4822 Disable the Correctable Errors Collector,
4823 see CONFIG_RAS_CEC help text.
4825 rcu_nocbs[=cpu-list]
4826 [KNL] The optional argument is a cpu list,
4829 In kernels built with CONFIG_RCU_NOCB_CPU=y,
4830 enable the no-callback CPU mode, which prevents
4831 such CPUs' callbacks from being invoked in
4832 softirq context. Invocation of such CPUs' RCU
4833 callbacks will instead be offloaded to "rcuox/N"
4834 kthreads created for that purpose, where "x" is
4835 "p" for RCU-preempt, "s" for RCU-sched, and "g"
4836 for the kthreads that mediate grace periods; and
4837 "N" is the CPU number. This reduces OS jitter on
4838 the offloaded CPUs, which can be useful for HPC
4839 and real-time workloads. It can also improve
4840 energy efficiency for asymmetric multiprocessors.
4842 If a cpulist is passed as an argument, the specified
4843 list of CPUs is set to no-callback mode from boot.
4845 Otherwise, if the '=' sign and the cpulist
4846 arguments are omitted, no CPU will be set to
4847 no-callback mode from boot but the mode may be
4848 toggled at runtime via cpusets.
4850 Note that this argument takes precedence over
4851 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
4854 Rather than requiring that offloaded CPUs
4855 (specified by rcu_nocbs= above) explicitly
4856 awaken the corresponding "rcuoN" kthreads,
4857 make these kthreads poll for callbacks.
4858 This improves the real-time response for the
4859 offloaded CPUs by relieving them of the need to
4860 wake up the corresponding kthread, but degrades
4861 energy efficiency by requiring that the kthreads
4862 periodically wake up to do the polling.
4864 rcutree.blimit= [KNL]
4865 Set maximum number of finished RCU callbacks to
4866 process in one batch.
4868 rcutree.do_rcu_barrier= [KNL]
4869 Request a call to rcu_barrier(). This is
4870 throttled so that userspace tests can safely
4871 hammer on the sysfs variable if they so choose.
4872 If triggered before the RCU grace-period machinery
4873 is fully active, this will error out with EAGAIN.
4875 rcutree.dump_tree= [KNL]
4876 Dump the structure of the rcu_node combining tree
4877 out at early boot. This is used for diagnostic
4878 purposes, to verify correct tree setup.
4880 rcutree.gp_cleanup_delay= [KNL]
4881 Set the number of jiffies to delay each step of
4882 RCU grace-period cleanup.
4884 rcutree.gp_init_delay= [KNL]
4885 Set the number of jiffies to delay each step of
4886 RCU grace-period initialization.
4888 rcutree.gp_preinit_delay= [KNL]
4889 Set the number of jiffies to delay each step of
4890 RCU grace-period pre-initialization, that is,
4891 the propagation of recent CPU-hotplug changes up
4892 the rcu_node combining tree.
4894 rcutree.jiffies_till_first_fqs= [KNL]
4895 Set delay from grace-period initialization to
4896 first attempt to force quiescent states.
4897 Units are jiffies, minimum value is zero,
4898 and maximum value is HZ.
4900 rcutree.jiffies_till_next_fqs= [KNL]
4901 Set delay between subsequent attempts to force
4902 quiescent states. Units are jiffies, minimum
4903 value is one, and maximum value is HZ.
4905 rcutree.jiffies_till_sched_qs= [KNL]
4906 Set required age in jiffies for a
4907 given grace period before RCU starts
4908 soliciting quiescent-state help from
4909 rcu_note_context_switch() and cond_resched().
4910 If not specified, the kernel will calculate
4911 a value based on the most recent settings
4912 of rcutree.jiffies_till_first_fqs
4913 and rcutree.jiffies_till_next_fqs.
4914 This calculated value may be viewed in
4915 rcutree.jiffies_to_sched_qs. Any attempt to set
4916 rcutree.jiffies_to_sched_qs will be cheerfully
4919 rcutree.kthread_prio= [KNL,BOOT]
4920 Set the SCHED_FIFO priority of the RCU per-CPU
4921 kthreads (rcuc/N). This value is also used for
4922 the priority of the RCU boost threads (rcub/N)
4923 and for the RCU grace-period kthreads (rcu_bh,
4924 rcu_preempt, and rcu_sched). If RCU_BOOST is
4925 set, valid values are 1-99 and the default is 1
4926 (the least-favored priority). Otherwise, when
4927 RCU_BOOST is not set, valid values are 0-99 and
4928 the default is zero (non-realtime operation).
4929 When RCU_NOCB_CPU is set, also adjust the
4930 priority of NOCB callback kthreads.
4932 rcutree.nocb_nobypass_lim_per_jiffy= [KNL]
4933 On callback-offloaded (rcu_nocbs) CPUs,
4934 RCU reduces the lock contention that would
4935 otherwise be caused by callback floods through
4936 use of the ->nocb_bypass list. However, in the
4937 common non-flooded case, RCU queues directly to
4938 the main ->cblist in order to avoid the extra
4939 overhead of the ->nocb_bypass list and its lock.
4940 But if there are too many callbacks queued during
4941 a single jiffy, RCU pre-queues the callbacks into
4942 the ->nocb_bypass queue. The definition of "too
4943 many" is supplied by this kernel boot parameter.
4945 rcutree.qhimark= [KNL]
4946 Set threshold of queued RCU callbacks beyond which
4947 batch limiting is disabled.
4949 rcutree.qlowmark= [KNL]
4950 Set threshold of queued RCU callbacks below which
4951 batch limiting is re-enabled.
4953 rcutree.qovld= [KNL]
4954 Set threshold of queued RCU callbacks beyond which
4955 RCU's force-quiescent-state scan will aggressively
4956 enlist help from cond_resched() and sched IPIs to
4957 help CPUs more quickly reach quiescent states.
4958 Set to less than zero to make this be set based
4959 on rcutree.qhimark at boot time and to zero to
4960 disable more aggressive help enlistment.
4962 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4963 Set the page-cache refill delay (in milliseconds)
4964 in response to low-memory conditions. The range
4965 of permitted values is in the range 0:100000.
4967 rcutree.rcu_divisor= [KNL]
4968 Set the shift-right count to use to compute
4969 the callback-invocation batch limit bl from
4970 the number of callbacks queued on this CPU.
4971 The result will be bounded below by the value of
4972 the rcutree.blimit kernel parameter. Every bl
4973 callbacks, the softirq handler will exit in
4974 order to allow the CPU to do other work.
4976 Please note that this callback-invocation batch
4977 limit applies only to non-offloaded callback
4978 invocation. Offloaded callbacks are instead
4979 invoked in the context of an rcuoc kthread, which
4980 scheduler will preempt as it does any other task.
4982 rcutree.rcu_fanout_exact= [KNL]
4983 Disable autobalancing of the rcu_node combining
4984 tree. This is used by rcutorture, and might
4985 possibly be useful for architectures having high
4986 cache-to-cache transfer latencies.
4988 rcutree.rcu_fanout_leaf= [KNL]
4989 Change the number of CPUs assigned to each
4990 leaf rcu_node structure. Useful for very
4991 large systems, which will choose the value 64,
4992 and for NUMA systems with large remote-access
4993 latencies, which will choose a value aligned
4994 with the appropriate hardware boundaries.
4996 rcutree.rcu_min_cached_objs= [KNL]
4997 Minimum number of objects which are cached and
4998 maintained per one CPU. Object size is equal
4999 to PAGE_SIZE. The cache allows to reduce the
5000 pressure to page allocator, also it makes the
5001 whole algorithm to behave better in low memory
5004 rcutree.rcu_nocb_gp_stride= [KNL]
5005 Set the number of NOCB callback kthreads in
5006 each group, which defaults to the square root
5007 of the number of CPUs. Larger numbers reduce
5008 the wakeup overhead on the global grace-period
5009 kthread, but increases that same overhead on
5010 each group's NOCB grace-period kthread.
5012 rcutree.rcu_kick_kthreads= [KNL]
5013 Cause the grace-period kthread to get an extra
5014 wake_up() if it sleeps three times longer than
5015 it should at force-quiescent-state time.
5016 This wake_up() will be accompanied by a
5017 WARN_ONCE() splat and an ftrace_dump().
5019 rcutree.rcu_resched_ns= [KNL]
5020 Limit the time spend invoking a batch of RCU
5021 callbacks to the specified number of nanoseconds.
5022 By default, this limit is checked only once
5023 every 32 callbacks in order to limit the pain
5024 inflicted by local_clock() overhead.
5026 rcutree.rcu_unlock_delay= [KNL]
5027 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
5028 this specifies an rcu_read_unlock()-time delay
5029 in microseconds. This defaults to zero.
5030 Larger delays increase the probability of
5031 catching RCU pointer leaks, that is, buggy use
5032 of RCU-protected pointers after the relevant
5033 rcu_read_unlock() has completed.
5035 rcutree.sysrq_rcu= [KNL]
5036 Commandeer a sysrq key to dump out Tree RCU's
5037 rcu_node tree with an eye towards determining
5038 why a new grace period has not yet started.
5040 rcutree.use_softirq= [KNL]
5041 If set to zero, move all RCU_SOFTIRQ processing to
5042 per-CPU rcuc kthreads. Defaults to a non-zero
5043 value, meaning that RCU_SOFTIRQ is used by default.
5044 Specify rcutree.use_softirq=0 to use rcuc kthreads.
5046 But note that CONFIG_PREEMPT_RT=y kernels disable
5047 this kernel boot parameter, forcibly setting it
5050 rcutree.enable_rcu_lazy= [KNL]
5051 To save power, batch RCU callbacks and flush after
5052 delay, memory pressure or callback list growing too
5055 rcuscale.gp_async= [KNL]
5056 Measure performance of asynchronous
5057 grace-period primitives such as call_rcu().
5059 rcuscale.gp_async_max= [KNL]
5060 Specify the maximum number of outstanding
5061 callbacks per writer thread. When a writer
5062 thread exceeds this limit, it invokes the
5063 corresponding flavor of rcu_barrier() to allow
5064 previously posted callbacks to drain.
5066 rcuscale.gp_exp= [KNL]
5067 Measure performance of expedited synchronous
5068 grace-period primitives.
5070 rcuscale.holdoff= [KNL]
5071 Set test-start holdoff period. The purpose of
5072 this parameter is to delay the start of the
5073 test until boot completes in order to avoid
5076 rcuscale.kfree_by_call_rcu= [KNL]
5077 In kernels built with CONFIG_RCU_LAZY=y, test
5078 call_rcu() instead of kfree_rcu().
5080 rcuscale.kfree_mult= [KNL]
5081 Instead of allocating an object of size kfree_obj,
5082 allocate one of kfree_mult * sizeof(kfree_obj).
5085 rcuscale.kfree_rcu_test= [KNL]
5086 Set to measure performance of kfree_rcu() flooding.
5088 rcuscale.kfree_rcu_test_double= [KNL]
5089 Test the double-argument variant of kfree_rcu().
5090 If this parameter has the same value as
5091 rcuscale.kfree_rcu_test_single, both the single-
5092 and double-argument variants are tested.
5094 rcuscale.kfree_rcu_test_single= [KNL]
5095 Test the single-argument variant of kfree_rcu().
5096 If this parameter has the same value as
5097 rcuscale.kfree_rcu_test_double, both the single-
5098 and double-argument variants are tested.
5100 rcuscale.kfree_nthreads= [KNL]
5101 The number of threads running loops of kfree_rcu().
5103 rcuscale.kfree_alloc_num= [KNL]
5104 Number of allocations and frees done in an iteration.
5106 rcuscale.kfree_loops= [KNL]
5107 Number of loops doing rcuscale.kfree_alloc_num number
5108 of allocations and frees.
5110 rcuscale.minruntime= [KNL]
5111 Set the minimum test run time in seconds. This
5112 does not affect the data-collection interval,
5113 but instead allows better measurement of things
5114 like CPU consumption.
5116 rcuscale.nreaders= [KNL]
5117 Set number of RCU readers. The value -1 selects
5118 N, where N is the number of CPUs. A value
5119 "n" less than -1 selects N-n+1, where N is again
5120 the number of CPUs. For example, -2 selects N
5121 (the number of CPUs), -3 selects N+1, and so on.
5122 A value of "n" less than or equal to -N selects
5125 rcuscale.nwriters= [KNL]
5126 Set number of RCU writers. The values operate
5127 the same as for rcuscale.nreaders.
5128 N, where N is the number of CPUs
5130 rcuscale.scale_type= [KNL]
5131 Specify the RCU implementation to test.
5133 rcuscale.shutdown= [KNL]
5134 Shut the system down after performance tests
5135 complete. This is useful for hands-off automated
5138 rcuscale.verbose= [KNL]
5139 Enable additional printk() statements.
5141 rcuscale.writer_holdoff= [KNL]
5142 Write-side holdoff between grace periods,
5143 in microseconds. The default of zero says
5146 rcuscale.writer_holdoff_jiffies= [KNL]
5147 Additional write-side holdoff between grace
5148 periods, but in jiffies. The default of zero
5151 rcutorture.fqs_duration= [KNL]
5152 Set duration of force_quiescent_state bursts
5155 rcutorture.fqs_holdoff= [KNL]
5156 Set holdoff time within force_quiescent_state bursts
5159 rcutorture.fqs_stutter= [KNL]
5160 Set wait time between force_quiescent_state bursts
5163 rcutorture.fwd_progress= [KNL]
5164 Specifies the number of kthreads to be used
5165 for RCU grace-period forward-progress testing
5166 for the types of RCU supporting this notion.
5167 Defaults to 1 kthread, values less than zero or
5168 greater than the number of CPUs cause the number
5171 rcutorture.fwd_progress_div= [KNL]
5172 Specify the fraction of a CPU-stall-warning
5173 period to do tight-loop forward-progress testing.
5175 rcutorture.fwd_progress_holdoff= [KNL]
5176 Number of seconds to wait between successive
5177 forward-progress tests.
5179 rcutorture.fwd_progress_need_resched= [KNL]
5180 Enclose cond_resched() calls within checks for
5181 need_resched() during tight-loop forward-progress
5184 rcutorture.gp_cond= [KNL]
5185 Use conditional/asynchronous update-side
5186 primitives, if available.
5188 rcutorture.gp_exp= [KNL]
5189 Use expedited update-side primitives, if available.
5191 rcutorture.gp_normal= [KNL]
5192 Use normal (non-expedited) asynchronous
5193 update-side primitives, if available.
5195 rcutorture.gp_sync= [KNL]
5196 Use normal (non-expedited) synchronous
5197 update-side primitives, if available. If all
5198 of rcutorture.gp_cond=, rcutorture.gp_exp=,
5199 rcutorture.gp_normal=, and rcutorture.gp_sync=
5200 are zero, rcutorture acts as if is interpreted
5201 they are all non-zero.
5203 rcutorture.irqreader= [KNL]
5204 Run RCU readers from irq handlers, or, more
5205 accurately, from a timer handler. Not all RCU
5206 flavors take kindly to this sort of thing.
5208 rcutorture.leakpointer= [KNL]
5209 Leak an RCU-protected pointer out of the reader.
5210 This can of course result in splats, and is
5211 intended to test the ability of things like
5212 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
5215 rcutorture.n_barrier_cbs= [KNL]
5216 Set callbacks/threads for rcu_barrier() testing.
5218 rcutorture.nfakewriters= [KNL]
5219 Set number of concurrent RCU writers. These just
5220 stress RCU, they don't participate in the actual
5221 test, hence the "fake".
5223 rcutorture.nocbs_nthreads= [KNL]
5224 Set number of RCU callback-offload togglers.
5225 Zero (the default) disables toggling.
5227 rcutorture.nocbs_toggle= [KNL]
5228 Set the delay in milliseconds between successive
5229 callback-offload toggling attempts.
5231 rcutorture.nreaders= [KNL]
5232 Set number of RCU readers. The value -1 selects
5233 N-1, where N is the number of CPUs. A value
5234 "n" less than -1 selects N-n-2, where N is again
5235 the number of CPUs. For example, -2 selects N
5236 (the number of CPUs), -3 selects N+1, and so on.
5238 rcutorture.object_debug= [KNL]
5239 Enable debug-object double-call_rcu() testing.
5241 rcutorture.onoff_holdoff= [KNL]
5242 Set time (s) after boot for CPU-hotplug testing.
5244 rcutorture.onoff_interval= [KNL]
5245 Set time (jiffies) between CPU-hotplug operations,
5246 or zero to disable CPU-hotplug testing.
5248 rcutorture.read_exit= [KNL]
5249 Set the number of read-then-exit kthreads used
5250 to test the interaction of RCU updaters and
5251 task-exit processing.
5253 rcutorture.read_exit_burst= [KNL]
5254 The number of times in a given read-then-exit
5255 episode that a set of read-then-exit kthreads
5258 rcutorture.read_exit_delay= [KNL]
5259 The delay, in seconds, between successive
5260 read-then-exit testing episodes.
5262 rcutorture.shuffle_interval= [KNL]
5263 Set task-shuffle interval (s). Shuffling tasks
5264 allows some CPUs to go into dyntick-idle mode
5265 during the rcutorture test.
5267 rcutorture.shutdown_secs= [KNL]
5268 Set time (s) after boot system shutdown. This
5269 is useful for hands-off automated testing.
5271 rcutorture.stall_cpu= [KNL]
5272 Duration of CPU stall (s) to test RCU CPU stall
5273 warnings, zero to disable.
5275 rcutorture.stall_cpu_block= [KNL]
5276 Sleep while stalling if set. This will result
5277 in warnings from preemptible RCU in addition to
5278 any other stall-related activity. Note that
5279 in kernels built with CONFIG_PREEMPTION=n and
5280 CONFIG_PREEMPT_COUNT=y, this parameter will
5281 cause the CPU to pass through a quiescent state.
5282 Given CONFIG_PREEMPTION=n, this will suppress
5283 RCU CPU stall warnings, but will instead result
5284 in scheduling-while-atomic splats.
5286 Use of this module parameter results in splats.
5289 rcutorture.stall_cpu_holdoff= [KNL]
5290 Time to wait (s) after boot before inducing stall.
5292 rcutorture.stall_cpu_irqsoff= [KNL]
5293 Disable interrupts while stalling if set.
5295 rcutorture.stall_gp_kthread= [KNL]
5296 Duration (s) of forced sleep within RCU
5297 grace-period kthread to test RCU CPU stall
5298 warnings, zero to disable. If both stall_cpu
5299 and stall_gp_kthread are specified, the
5300 kthread is starved first, then the CPU.
5302 rcutorture.stat_interval= [KNL]
5303 Time (s) between statistics printk()s.
5305 rcutorture.stutter= [KNL]
5306 Time (s) to stutter testing, for example, specifying
5307 five seconds causes the test to run for five seconds,
5308 wait for five seconds, and so on. This tests RCU's
5309 ability to transition abruptly to and from idle.
5311 rcutorture.test_boost= [KNL]
5312 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
5313 "Maybe" means test if the RCU implementation
5314 under test support RCU priority boosting.
5316 rcutorture.test_boost_duration= [KNL]
5317 Duration (s) of each individual boost test.
5319 rcutorture.test_boost_interval= [KNL]
5320 Interval (s) between each boost test.
5322 rcutorture.test_no_idle_hz= [KNL]
5323 Test RCU's dyntick-idle handling. See also the
5324 rcutorture.shuffle_interval parameter.
5326 rcutorture.torture_type= [KNL]
5327 Specify the RCU implementation to test.
5329 rcutorture.verbose= [KNL]
5330 Enable additional printk() statements.
5332 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
5333 Dump ftrace buffer after reporting RCU CPU
5336 rcupdate.rcu_cpu_stall_notifiers= [KNL]
5337 Provide RCU CPU stall notifiers, but see the
5338 warnings in the RCU_CPU_STALL_NOTIFIER Kconfig
5339 option's help text. TL;DR: You almost certainly
5340 do not want rcupdate.rcu_cpu_stall_notifiers.
5342 rcupdate.rcu_cpu_stall_suppress= [KNL]
5343 Suppress RCU CPU stall warning messages.
5345 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
5346 Suppress RCU CPU stall warning messages and
5347 rcutorture writer stall warnings that occur
5348 during early boot, that is, during the time
5349 before the init task is spawned.
5351 rcupdate.rcu_cpu_stall_timeout= [KNL]
5352 Set timeout for RCU CPU stall warning messages.
5353 The value is in seconds and the maximum allowed
5354 value is 300 seconds.
5356 rcupdate.rcu_exp_cpu_stall_timeout= [KNL]
5357 Set timeout for expedited RCU CPU stall warning
5358 messages. The value is in milliseconds
5359 and the maximum allowed value is 21000
5360 milliseconds. Please note that this value is
5361 adjusted to an arch timer tick resolution.
5362 Setting this to zero causes the value from
5363 rcupdate.rcu_cpu_stall_timeout to be used (after
5364 conversion from seconds to milliseconds).
5366 rcupdate.rcu_cpu_stall_cputime= [KNL]
5367 Provide statistics on the cputime and count of
5368 interrupts and tasks during the sampling period. For
5369 multiple continuous RCU stalls, all sampling periods
5370 begin at half of the first RCU stall timeout.
5372 rcupdate.rcu_exp_stall_task_details= [KNL]
5373 Print stack dumps of any tasks blocking the
5374 current expedited RCU grace period during an
5375 expedited RCU CPU stall warning.
5377 rcupdate.rcu_expedited= [KNL]
5378 Use expedited grace-period primitives, for
5379 example, synchronize_rcu_expedited() instead
5380 of synchronize_rcu(). This reduces latency,
5381 but can increase CPU utilization, degrade
5382 real-time latency, and degrade energy efficiency.
5383 No effect on CONFIG_TINY_RCU kernels.
5385 rcupdate.rcu_normal= [KNL]
5386 Use only normal grace-period primitives,
5387 for example, synchronize_rcu() instead of
5388 synchronize_rcu_expedited(). This improves
5389 real-time latency, CPU utilization, and
5390 energy efficiency, but can expose users to
5391 increased grace-period latency. This parameter
5392 overrides rcupdate.rcu_expedited. No effect on
5393 CONFIG_TINY_RCU kernels.
5395 rcupdate.rcu_normal_after_boot= [KNL]
5396 Once boot has completed (that is, after
5397 rcu_end_inkernel_boot() has been invoked), use
5398 only normal grace-period primitives. No effect
5399 on CONFIG_TINY_RCU kernels.
5401 But note that CONFIG_PREEMPT_RT=y kernels enables
5402 this kernel boot parameter, forcibly setting
5403 it to the value one, that is, converting any
5404 post-boot attempt at an expedited RCU grace
5405 period to instead use normal non-expedited
5406 grace-period processing.
5408 rcupdate.rcu_task_collapse_lim= [KNL]
5409 Set the maximum number of callbacks present
5410 at the beginning of a grace period that allows
5411 the RCU Tasks flavors to collapse back to using
5412 a single callback queue. This switching only
5413 occurs when rcupdate.rcu_task_enqueue_lim is
5414 set to the default value of -1.
5416 rcupdate.rcu_task_contend_lim= [KNL]
5417 Set the minimum number of callback-queuing-time
5418 lock-contention events per jiffy required to
5419 cause the RCU Tasks flavors to switch to per-CPU
5420 callback queuing. This switching only occurs
5421 when rcupdate.rcu_task_enqueue_lim is set to
5422 the default value of -1.
5424 rcupdate.rcu_task_enqueue_lim= [KNL]
5425 Set the number of callback queues to use for the
5426 RCU Tasks family of RCU flavors. The default
5427 of -1 allows this to be automatically (and
5428 dynamically) adjusted. This parameter is intended
5431 rcupdate.rcu_task_ipi_delay= [KNL]
5432 Set time in jiffies during which RCU tasks will
5433 avoid sending IPIs, starting with the beginning
5434 of a given grace period. Setting a large
5435 number avoids disturbing real-time workloads,
5436 but lengthens grace periods.
5438 rcupdate.rcu_task_lazy_lim= [KNL]
5439 Number of callbacks on a given CPU that will
5440 cancel laziness on that CPU. Use -1 to disable
5441 cancellation of laziness, but be advised that
5442 doing so increases the danger of OOM due to
5445 rcupdate.rcu_task_stall_info= [KNL]
5446 Set initial timeout in jiffies for RCU task stall
5447 informational messages, which give some indication
5448 of the problem for those not patient enough to
5449 wait for ten minutes. Informational messages are
5450 only printed prior to the stall-warning message
5451 for a given grace period. Disable with a value
5452 less than or equal to zero. Defaults to ten
5453 seconds. A change in value does not take effect
5454 until the beginning of the next grace period.
5456 rcupdate.rcu_task_stall_info_mult= [KNL]
5457 Multiplier for time interval between successive
5458 RCU task stall informational messages for a given
5459 RCU tasks grace period. This value is clamped
5460 to one through ten, inclusive. It defaults to
5461 the value three, so that the first informational
5462 message is printed 10 seconds into the grace
5463 period, the second at 40 seconds, the third at
5464 160 seconds, and then the stall warning at 600
5465 seconds would prevent a fourth at 640 seconds.
5467 rcupdate.rcu_task_stall_timeout= [KNL]
5468 Set timeout in jiffies for RCU task stall
5469 warning messages. Disable with a value less
5470 than or equal to zero. Defaults to ten minutes.
5471 A change in value does not take effect until
5472 the beginning of the next grace period.
5474 rcupdate.rcu_tasks_lazy_ms= [KNL]
5475 Set timeout in milliseconds RCU Tasks asynchronous
5476 callback batching for call_rcu_tasks().
5477 A negative value will take the default. A value
5478 of zero will disable batching. Batching is
5479 always disabled for synchronize_rcu_tasks().
5481 rcupdate.rcu_tasks_rude_lazy_ms= [KNL]
5482 Set timeout in milliseconds RCU Tasks
5483 Rude asynchronous callback batching for
5484 call_rcu_tasks_rude(). A negative value
5485 will take the default. A value of zero will
5486 disable batching. Batching is always disabled
5487 for synchronize_rcu_tasks_rude().
5489 rcupdate.rcu_tasks_trace_lazy_ms= [KNL]
5490 Set timeout in milliseconds RCU Tasks
5491 Trace asynchronous callback batching for
5492 call_rcu_tasks_trace(). A negative value
5493 will take the default. A value of zero will
5494 disable batching. Batching is always disabled
5495 for synchronize_rcu_tasks_trace().
5497 rcupdate.rcu_self_test= [KNL]
5498 Run the RCU early boot self tests
5502 Run specified binary instead of /init from the ramdisk,
5503 used for early userspace startup. See initrd.
5506 force - Override the decision by the kernel to hide the
5507 advertisement of RDRAND support (this affects
5508 certain AMD processors because of buggy BIOS
5509 support, specifically around the suspend/resume
5513 Turn on/off individual RDT features. List is:
5514 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
5516 E.g. to turn on cmt and turn off mba use:
5520 Format (x86 or x86_64):
5521 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
5523 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
5525 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
5526 (prefix with 'panic_' to set mode for panic
5528 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
5529 reboot_force is either force or not specified,
5530 reboot_cpu is s[mp]#### with #### being the processor
5531 to be used for rebooting.
5533 refscale.holdoff= [KNL]
5534 Set test-start holdoff period. The purpose of
5535 this parameter is to delay the start of the
5536 test until boot completes in order to avoid
5539 refscale.lookup_instances= [KNL]
5540 Number of data elements to use for the forms of
5541 SLAB_TYPESAFE_BY_RCU testing. A negative number
5542 is negated and multiplied by nr_cpu_ids, while
5543 zero specifies nr_cpu_ids.
5545 refscale.loops= [KNL]
5546 Set the number of loops over the synchronization
5547 primitive under test. Increasing this number
5548 reduces noise due to loop start/end overhead,
5549 but the default has already reduced the per-pass
5550 noise to a handful of picoseconds on ca. 2020
5553 refscale.nreaders= [KNL]
5554 Set number of readers. The default value of -1
5555 selects N, where N is roughly 75% of the number
5556 of CPUs. A value of zero is an interesting choice.
5558 refscale.nruns= [KNL]
5559 Set number of runs, each of which is dumped onto
5562 refscale.readdelay= [KNL]
5563 Set the read-side critical-section duration,
5564 measured in microseconds.
5566 refscale.scale_type= [KNL]
5567 Specify the read-protection implementation to test.
5569 refscale.shutdown= [KNL]
5570 Shut down the system at the end of the performance
5571 test. This defaults to 1 (shut it down) when
5572 refscale is built into the kernel and to 0 (leave
5573 it running) when refscale is built as a module.
5575 refscale.verbose= [KNL]
5576 Enable additional printk() statements.
5578 refscale.verbose_batched= [KNL]
5579 Batch the additional printk() statements. If zero
5580 (the default) or negative, print everything. Otherwise,
5581 print every Nth verbose statement, where N is the value
5584 regulator_ignore_unused
5586 Prevents regulator framework from disabling regulators
5587 that are unused, due no driver claiming them. This may
5588 be useful for debug and development, but should not be
5589 needed on a platform with proper driver support.
5592 [KNL, SMP] Set scheduler's default relax_domain_level.
5593 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
5595 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
5596 Format: <base1>,<size1>[,<base2>,<size2>,...]
5597 Reserve I/O ports or memory so the kernel won't use
5598 them. If <base> is less than 0x10000, the region
5599 is assumed to be I/O ports; otherwise it is memory.
5601 reservetop= [X86-32,EARLY]
5603 Reserves a hole at the top of the kernel virtual
5606 reset_devices [KNL] Force drivers to reset the underlying device
5607 during initialization.
5610 Specify the partition device for software suspend
5612 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
5614 resume_offset= [SWSUSP]
5615 Specify the offset from the beginning of the partition
5616 given by "resume=" at which the swap header is located,
5617 in <PAGE_SIZE> units (needed only for swap files).
5618 See Documentation/power/swsusp-and-swap-files.rst
5620 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
5621 read the resume files
5623 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
5624 Useful for devices that are detected asynchronously
5625 (e.g. USB and MMC devices).
5627 retain_initrd [RAM] Keep initrd memory after extraction. After boot, it will
5628 be accessible via /sys/firmware/initrd.
5630 retbleed= [X86] Control mitigation of RETBleed (Arbitrary
5631 Speculative Code Execution with Return Instructions)
5634 AMD-based UNRET and IBPB mitigations alone do not stop
5635 sibling threads from influencing the predictions of other
5636 sibling threads. For that reason, STIBP is used on pro-
5637 cessors that support it, and mitigate SMT on processors
5641 auto - automatically select a migitation
5642 auto,nosmt - automatically select a mitigation,
5643 disabling SMT if necessary for
5644 the full mitigation (only on Zen1
5645 and older without STIBP).
5646 ibpb - On AMD, mitigate short speculation
5647 windows on basic block boundaries too.
5648 Safe, highest perf impact. It also
5649 enables STIBP if present. Not suitable
5651 ibpb,nosmt - Like "ibpb" above but will disable SMT
5652 when STIBP is not available. This is
5653 the alternative for systems which do not
5655 unret - Force enable untrained return thunks,
5656 only effective on AMD f15h-f17h based
5658 unret,nosmt - Like unret, but will disable SMT when STIBP
5659 is not available. This is the alternative for
5660 systems which do not have STIBP.
5662 Selecting 'auto' will choose a mitigation method at run
5663 time according to the CPU.
5665 Not specifying this option is equivalent to retbleed=auto.
5667 rfkill.default_state=
5668 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5669 etc. communication is blocked by default.
5672 rfkill.master_switch_mode=
5673 0 The "airplane mode" button does nothing.
5674 1 The "airplane mode" button toggles between everything
5675 blocked and the previous configuration.
5676 2 The "airplane mode" button toggles between everything
5677 blocked and everything unblocked.
5679 rhash_entries= [KNL,NET]
5680 Set number of hash buckets for route cache
5683 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5686 riscv_isa_fallback [RISCV,EARLY]
5687 When CONFIG_RISCV_ISA_FALLBACK is not enabled, permit
5688 falling back to detecting extension support by parsing
5689 "riscv,isa" property on devicetree systems when the
5690 replacement properties are not found. See the Kconfig
5691 entry for RISCV_ISA_FALLBACK.
5693 ro [KNL] Mount root device read-only on boot
5696 on Mark read-only kernel memory as read-only (default).
5697 off Leave read-only kernel memory writable for debugging.
5698 full Mark read-only kernel memory and aliases as read-only
5703 Enable the uart passthrough on the designated usb port
5704 on Rockchip SoCs. When active, the signals of the
5705 debug-uart get routed to the D+ and D- pins of the usb
5706 port and the regular usb controller gets disabled.
5708 root= [KNL] Root filesystem
5709 Usually this a a block device specifier of some kind,
5710 see the early_lookup_bdev comment in
5711 block/early-lookup.c for details.
5712 Alternatively this can be "ram" for the legacy initial
5713 ramdisk, "nfs" and "cifs" for root on a network file
5714 system, or "mtd" and "ubi" for mounting from raw flash.
5716 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5717 mount the root filesystem
5719 rootflags= [KNL] Set root filesystem mount option string
5721 rootfstype= [KNL] Set root filesystem type
5723 rootwait [KNL] Wait (indefinitely) for root device to show up.
5724 Useful for devices that are detected asynchronously
5725 (e.g. USB and MMC devices).
5727 rootwait= [KNL] Maximum time (in seconds) to wait for root device
5728 to show up before attempting to mount the root
5731 rproc_mem=nn[KMG][@address]
5732 [KNL,ARM,CMA] Remoteproc physical memory block.
5733 Memory area to be used by remote processor image,
5736 rw [KNL] Mount root device read-write on boot
5738 S [KNL] Run init in single mode
5740 s390_iommu= [HW,S390]
5741 Set s390 IOTLB flushing mode
5743 With strict flushing every unmap operation will result
5744 in an IOTLB flush. Default is lazy flushing before
5745 reuse, which is faster. Deprecated, equivalent to
5748 s390_iommu_aperture= [KNL,S390]
5749 Specifies the size of the per device DMA address space
5750 accessible through the DMA and IOMMU APIs as a decimal
5751 factor of the size of main memory.
5752 The default is 1 meaning that one can concurrently use
5753 as many DMA addresses as physical memory is installed,
5754 if supported by hardware, and thus map all of memory
5755 once. With a value of 2 one can map all of memory twice
5756 and so on. As a special case a factor of 0 imposes no
5757 restrictions other than those given by hardware at the
5758 cost of significant additional memory use for tables.
5761 See drivers/net/irda/sa1100_ir.c.
5763 sched_verbose [KNL,EARLY] Enables verbose scheduler debug messages.
5765 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5766 Allowed values are enable and disable. This feature
5767 incurs a small amount of overhead in the scheduler
5768 but is useful for debugging and performance tuning.
5770 sched_thermal_decay_shift=
5771 [KNL, SMP] Set a decay shift for scheduler thermal
5772 pressure signal. Thermal pressure signal follows the
5773 default decay period of other scheduler pelt
5774 signals(usually 32 ms but configurable). Setting
5775 sched_thermal_decay_shift will left shift the decay
5776 period for the thermal pressure signal by the shift
5778 i.e. with the default pelt decay period of 32 ms
5779 sched_thermal_decay_shift thermal pressure decay pr
5783 Format: integer between 0 and 10
5786 scftorture.holdoff= [KNL]
5787 Number of seconds to hold off before starting
5788 test. Defaults to zero for module insertion and
5789 to 10 seconds for built-in smp_call_function()
5792 scftorture.longwait= [KNL]
5793 Request ridiculously long waits randomly selected
5794 up to the chosen limit in seconds. Zero (the
5795 default) disables this feature. Please note
5796 that requesting even small non-zero numbers of
5797 seconds can result in RCU CPU stall warnings,
5798 softlockup complaints, and so on.
5800 scftorture.nthreads= [KNL]
5801 Number of kthreads to spawn to invoke the
5802 smp_call_function() family of functions.
5803 The default of -1 specifies a number of kthreads
5804 equal to the number of CPUs.
5806 scftorture.onoff_holdoff= [KNL]
5807 Number seconds to wait after the start of the
5808 test before initiating CPU-hotplug operations.
5810 scftorture.onoff_interval= [KNL]
5811 Number seconds to wait between successive
5812 CPU-hotplug operations. Specifying zero (which
5813 is the default) disables CPU-hotplug operations.
5815 scftorture.shutdown_secs= [KNL]
5816 The number of seconds following the start of the
5817 test after which to shut down the system. The
5818 default of zero avoids shutting down the system.
5819 Non-zero values are useful for automated tests.
5821 scftorture.stat_interval= [KNL]
5822 The number of seconds between outputting the
5823 current test statistics to the console. A value
5824 of zero disables statistics output.
5826 scftorture.stutter_cpus= [KNL]
5827 The number of jiffies to wait between each change
5828 to the set of CPUs under test.
5830 scftorture.use_cpus_read_lock= [KNL]
5831 Use use_cpus_read_lock() instead of the default
5832 preempt_disable() to disable CPU hotplug
5833 while invoking one of the smp_call_function*()
5836 scftorture.verbose= [KNL]
5837 Enable additional printk() statements.
5839 scftorture.weight_single= [KNL]
5840 The probability weighting to use for the
5841 smp_call_function_single() function with a zero
5842 "wait" parameter. A value of -1 selects the
5843 default if all other weights are -1. However,
5844 if at least one weight has some other value, a
5845 value of -1 will instead select a weight of zero.
5847 scftorture.weight_single_wait= [KNL]
5848 The probability weighting to use for the
5849 smp_call_function_single() function with a
5850 non-zero "wait" parameter. See weight_single.
5852 scftorture.weight_many= [KNL]
5853 The probability weighting to use for the
5854 smp_call_function_many() function with a zero
5855 "wait" parameter. See weight_single.
5856 Note well that setting a high probability for
5857 this weighting can place serious IPI load
5860 scftorture.weight_many_wait= [KNL]
5861 The probability weighting to use for the
5862 smp_call_function_many() function with a
5863 non-zero "wait" parameter. See weight_single
5866 scftorture.weight_all= [KNL]
5867 The probability weighting to use for the
5868 smp_call_function_all() function with a zero
5869 "wait" parameter. See weight_single and
5872 scftorture.weight_all_wait= [KNL]
5873 The probability weighting to use for the
5874 smp_call_function_all() function with a
5875 non-zero "wait" parameter. See weight_single
5878 skew_tick= [KNL,EARLY] Offset the periodic timer tick per cpu to mitigate
5879 xtime_lock contention on larger systems, and/or RCU lock
5880 contention on all systems with CONFIG_MAXSMP set.
5881 Format: { "0" | "1" }
5882 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5884 Note: increases power consumption, thus should only be
5885 enabled if running jitter sensitive (HPC/RT) workloads.
5887 security= [SECURITY] Choose a legacy "major" security module to
5888 enable at boot. This has been deprecated by the
5891 selinux= [SELINUX] Disable or enable SELinux at boot time.
5892 Format: { "0" | "1" }
5893 See security/selinux/Kconfig help text.
5898 serialnumber [BUGS=X86-32]
5900 sev=option[,option...] [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
5903 Maximal number of shapers.
5905 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
5906 Limit apic dumping. The parameter defines the maximal
5907 number of local apics being dumped. Also it is possible
5908 to set it to "all" by meaning -- no limit here.
5909 Format: { 1 (default) | 2 | ... | all }.
5910 The parameter valid if only apic=debug or
5911 apic=verbose is specified.
5912 Example: apic=debug show_lapic=all
5920 Enable merging of slabs with similar size when the
5921 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5924 Disable merging of slabs with similar size. May be
5925 necessary if there is some reason to distinguish
5926 allocs to different slabs, especially in hardened
5927 environments where the risk of heap overflows and
5928 layout control by attackers can usually be
5929 frustrated by disabling merging. This will reduce
5930 most of the exposure of a heap attack to a single
5931 cache (risks via metadata attacks are mostly
5932 unchanged). Debug options disable merging on their
5934 For more information see Documentation/mm/slub.rst.
5936 slab_max_order= [MM, SLAB]
5937 Determines the maximum allowed order for slabs.
5938 A high setting may cause OOMs due to memory
5939 fragmentation. Defaults to 1 for systems with
5940 more than 32MB of RAM, 0 otherwise.
5942 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5943 Enabling slub_debug allows one to determine the
5944 culprit if slab objects become corrupted. Enabling
5945 slub_debug can create guard zones around objects and
5946 may poison objects when not in use. Also tracks the
5947 last alloc / free. For more information see
5948 Documentation/mm/slub.rst.
5950 slub_max_order= [MM, SLUB]
5951 Determines the maximum allowed order for slabs.
5952 A high setting may cause OOMs due to memory
5953 fragmentation. For more information see
5954 Documentation/mm/slub.rst.
5956 slub_min_objects= [MM, SLUB]
5957 The minimum number of objects per slab. SLUB will
5958 increase the slab order up to slub_max_order to
5959 generate a sufficiently large slab able to contain
5960 the number of objects indicated. The higher the number
5961 of objects the smaller the overhead of tracking slabs
5962 and the less frequently locks need to be acquired.
5963 For more information see Documentation/mm/slub.rst.
5965 slub_min_order= [MM, SLUB]
5966 Determines the minimum page order for slabs. Must be
5967 lower than slub_max_order.
5968 For more information see Documentation/mm/slub.rst.
5970 slub_merge [MM, SLUB]
5971 Same with slab_merge.
5973 slub_nomerge [MM, SLUB]
5974 Same with slab_nomerge. This is supported for legacy.
5975 See slab_nomerge for more information.
5978 Format: <io1>[,<io2>[,...,<io8>]]
5980 smp.csd_lock_timeout= [KNL]
5981 Specify the period of time in milliseconds
5982 that smp_call_function() and friends will wait
5983 for a CPU to release the CSD lock. This is
5984 useful when diagnosing bugs involving CPUs
5985 disabling interrupts for extended periods
5986 of time. Defaults to 5,000 milliseconds, and
5987 setting a value of zero disables this feature.
5988 This feature may be more efficiently disabled
5989 using the csdlock_debug- kernel parameter.
5991 smp.panic_on_ipistall= [KNL]
5992 If a csd_lock_timeout extends for more than
5993 the specified number of milliseconds, panic the
5994 system. By default, let CSD-lock acquisition
5995 take as long as they take. Specifying 300,000
5996 for this value provides a 5-minute timeout.
5998 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5999 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
6000 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
6001 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
6002 smsc-ircc2.ircc_irq= [HW] IRQ line
6003 smsc-ircc2.ircc_dma= [HW] DMA channel
6004 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
6005 0: Toshiba Satellite 1800 (GP data pin select)
6006 1: Fast pin select (default)
6009 smt= [KNL,MIPS,S390,EARLY] Set the maximum number of threads
6010 (logical CPUs) to use per physical CPU on systems
6011 capable of symmetric multithreading (SMT). Will
6012 be capped to the actual hardware limit.
6014 Default: -1 (no limit)
6017 [KNL] Should the soft-lockup detector generate panics.
6020 A value of 1 instructs the soft-lockup detector
6021 to panic the machine when a soft-lockup occurs. It is
6022 also controlled by the kernel.softlockup_panic sysctl
6023 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
6024 respective build-time switch to that functionality.
6026 softlockup_all_cpu_backtrace=
6027 [KNL] Should the soft-lockup detector generate
6028 backtraces on all cpus.
6031 sonypi.*= [HW] Sony Programmable I/O Control Device driver
6032 See Documentation/admin-guide/laptops/sonypi.rst
6034 spectre_v2= [X86,EARLY] Control mitigation of Spectre variant 2
6035 (indirect branch speculation) vulnerability.
6036 The default operation protects the kernel from
6039 on - unconditionally enable, implies
6041 off - unconditionally disable, implies
6043 auto - kernel detects whether your CPU model is
6046 Selecting 'on' will, and 'auto' may, choose a
6047 mitigation method at run time according to the
6048 CPU, the available microcode, the setting of the
6049 CONFIG_RETPOLINE configuration option, and the
6050 compiler with which the kernel was built.
6052 Selecting 'on' will also enable the mitigation
6053 against user space to user space task attacks.
6055 Selecting 'off' will disable both the kernel and
6056 the user space protections.
6058 Specific mitigations can also be selected manually:
6060 retpoline - replace indirect branches
6061 retpoline,generic - Retpolines
6062 retpoline,lfence - LFENCE; indirect branch
6063 retpoline,amd - alias for retpoline,lfence
6064 eibrs - Enhanced/Auto IBRS
6065 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines
6066 eibrs,lfence - Enhanced/Auto IBRS + LFENCE
6067 ibrs - use IBRS to protect kernel
6069 Not specifying this option is equivalent to
6073 [X86] Control mitigation of Spectre variant 2
6074 (indirect branch speculation) vulnerability between
6077 on - Unconditionally enable mitigations. Is
6078 enforced by spectre_v2=on
6080 off - Unconditionally disable mitigations. Is
6081 enforced by spectre_v2=off
6083 prctl - Indirect branch speculation is enabled,
6084 but mitigation can be enabled via prctl
6085 per thread. The mitigation control state
6086 is inherited on fork.
6089 - Like "prctl" above, but only STIBP is
6090 controlled per thread. IBPB is issued
6091 always when switching between different user
6095 - Same as "prctl" above, but all seccomp
6096 threads will enable the mitigation unless
6097 they explicitly opt out.
6100 - Like "seccomp" above, but only STIBP is
6101 controlled per thread. IBPB is issued
6102 always when switching between different
6103 user space processes.
6105 auto - Kernel selects the mitigation depending on
6106 the available CPU features and vulnerability.
6108 Default mitigation: "prctl"
6110 Not specifying this option is equivalent to
6111 spectre_v2_user=auto.
6113 spec_rstack_overflow=
6114 [X86,EARLY] Control RAS overflow mitigation on AMD Zen CPUs
6116 off - Disable mitigation
6117 microcode - Enable microcode mitigation only
6118 safe-ret - Enable sw-only safe RET mitigation (default)
6119 ibpb - Enable mitigation by issuing IBPB on
6121 ibpb-vmexit - Issue IBPB only on VMEXIT
6122 (cloud-specific mitigation)
6124 spec_store_bypass_disable=
6125 [HW,EARLY] Control Speculative Store Bypass (SSB) Disable mitigation
6126 (Speculative Store Bypass vulnerability)
6128 Certain CPUs are vulnerable to an exploit against a
6129 a common industry wide performance optimization known
6130 as "Speculative Store Bypass" in which recent stores
6131 to the same memory location may not be observed by
6132 later loads during speculative execution. The idea
6133 is that such stores are unlikely and that they can
6134 be detected prior to instruction retirement at the
6135 end of a particular speculation execution window.
6137 In vulnerable processors, the speculatively forwarded
6138 store can be used in a cache side channel attack, for
6139 example to read memory to which the attacker does not
6140 directly have access (e.g. inside sandboxed code).
6142 This parameter controls whether the Speculative Store
6143 Bypass optimization is used.
6145 On x86 the options are:
6147 on - Unconditionally disable Speculative Store Bypass
6148 off - Unconditionally enable Speculative Store Bypass
6149 auto - Kernel detects whether the CPU model contains an
6150 implementation of Speculative Store Bypass and
6151 picks the most appropriate mitigation. If the
6152 CPU is not vulnerable, "off" is selected. If the
6153 CPU is vulnerable the default mitigation is
6154 architecture and Kconfig dependent. See below.
6155 prctl - Control Speculative Store Bypass per thread
6156 via prctl. Speculative Store Bypass is enabled
6157 for a process by default. The state of the control
6158 is inherited on fork.
6159 seccomp - Same as "prctl" above, but all seccomp threads
6160 will disable SSB unless they explicitly opt out.
6162 Default mitigations:
6165 On powerpc the options are:
6167 on,auto - On Power8 and Power9 insert a store-forwarding
6168 barrier on kernel entry and exit. On Power7
6169 perform a software flush on kernel entry and
6173 Not specifying this option is equivalent to
6174 spec_store_bypass_disable=auto.
6176 spia_io_base= [HW,MTD]
6182 [X86] Enable split lock detection or bus lock detection
6184 When enabled (and if hardware support is present), atomic
6185 instructions that access data across cache line
6186 boundaries will result in an alignment check exception
6187 for split lock detection or a debug exception for
6192 warn - the kernel will emit rate-limited warnings
6193 about applications triggering the #AC
6194 exception or the #DB exception. This mode is
6195 the default on CPUs that support split lock
6196 detection or bus lock detection. Default
6197 behavior is by #AC if both features are
6198 enabled in hardware.
6200 fatal - the kernel will send SIGBUS to applications
6201 that trigger the #AC exception or the #DB
6202 exception. Default behavior is by #AC if
6203 both features are enabled in hardware.
6206 Set system wide rate limit to N bus locks
6207 per second for bus lock detection.
6210 N/A for split lock detection.
6213 If an #AC exception is hit in the kernel or in
6214 firmware (i.e. not while executing in user mode)
6215 the kernel will oops in either "warn" or "fatal"
6218 #DB exception for bus lock is triggered only when
6221 srbds= [X86,INTEL,EARLY]
6222 Control the Special Register Buffer Data Sampling
6225 Certain CPUs are vulnerable to an MDS-like
6226 exploit which can leak bits from the random
6229 By default, this issue is mitigated by
6230 microcode. However, the microcode fix can cause
6231 the RDRAND and RDSEED instructions to become
6232 much slower. Among other effects, this will
6233 result in reduced throughput from /dev/urandom.
6235 The microcode mitigation can be disabled with
6236 the following option:
6238 off: Disable mitigation and remove
6239 performance impact to RDRAND and RDSEED
6241 srcutree.big_cpu_lim [KNL]
6242 Specifies the number of CPUs constituting a
6243 large system, such that srcu_struct structures
6244 should immediately allocate an srcu_node array.
6245 This kernel-boot parameter defaults to 128,
6246 but takes effect only when the low-order four
6247 bits of srcutree.convert_to_big is equal to 3
6250 srcutree.convert_to_big [KNL]
6251 Specifies under what conditions an SRCU tree
6252 srcu_struct structure will be converted to big
6253 form, that is, with an rcu_node tree:
6256 1: At init_srcu_struct() time.
6257 2: When rcutorture decides to.
6258 3: Decide at boot time (default).
6259 0x1X: Above plus if high contention.
6261 Either way, the srcu_node tree will be sized based
6262 on the actual runtime number of CPUs (nr_cpu_ids)
6263 instead of the compile-time CONFIG_NR_CPUS.
6265 srcutree.counter_wrap_check [KNL]
6266 Specifies how frequently to check for
6267 grace-period sequence counter wrap for the
6268 srcu_data structure's ->srcu_gp_seq_needed field.
6269 The greater the number of bits set in this kernel
6270 parameter, the less frequently counter wrap will
6271 be checked for. Note that the bottom two bits
6274 srcutree.exp_holdoff [KNL]
6275 Specifies how many nanoseconds must elapse
6276 since the end of the last SRCU grace period for
6277 a given srcu_struct until the next normal SRCU
6278 grace period will be considered for automatic
6279 expediting. Set to zero to disable automatic
6282 srcutree.srcu_max_nodelay [KNL]
6283 Specifies the number of no-delay instances
6284 per jiffy for which the SRCU grace period
6285 worker thread will be rescheduled with zero
6286 delay. Beyond this limit, worker thread will
6287 be rescheduled with a sleep delay of one jiffy.
6289 srcutree.srcu_max_nodelay_phase [KNL]
6290 Specifies the per-grace-period phase, number of
6291 non-sleeping polls of readers. Beyond this limit,
6292 grace period worker thread will be rescheduled
6293 with a sleep delay of one jiffy, between each
6294 rescan of the readers, for a grace period phase.
6296 srcutree.srcu_retry_check_delay [KNL]
6297 Specifies number of microseconds of non-sleeping
6298 delay between each non-sleeping poll of readers.
6300 srcutree.small_contention_lim [KNL]
6301 Specifies the number of update-side contention
6302 events per jiffy will be tolerated before
6303 initiating a conversion of an srcu_struct
6304 structure to big form. Note that the value of
6305 srcutree.convert_to_big must have the 0x10 bit
6306 set for contention-based conversions to occur.
6308 ssbd= [ARM64,HW,EARLY]
6309 Speculative Store Bypass Disable control
6311 On CPUs that are vulnerable to the Speculative
6312 Store Bypass vulnerability and offer a
6313 firmware based mitigation, this parameter
6314 indicates how the mitigation should be used:
6316 force-on: Unconditionally enable mitigation for
6317 for both kernel and userspace
6318 force-off: Unconditionally disable mitigation for
6319 for both kernel and userspace
6320 kernel: Always enable mitigation in the
6321 kernel, and offer a prctl interface
6322 to allow userspace to register its
6323 interest in being mitigated too.
6325 stack_guard_gap= [MM]
6326 override the default stack gap protection. The value
6327 is in page units and it defines how many pages prior
6328 to (for stacks growing down) resp. after (for stacks
6329 growing up) the main stack are reserved for no other
6330 mapping. Default value is 256 pages.
6332 stack_depot_disable= [KNL,EARLY]
6333 Setting this to true through kernel command line will
6334 disable the stack depot thereby saving the static memory
6335 consumed by the stack hash table. By default this is set
6339 Enabled the stack tracer on boot up.
6341 stacktrace_filter=[function-list]
6342 [FTRACE] Limit the functions that the stack tracer
6343 will trace at boot up. function-list is a comma-separated
6344 list of functions. This list can be changed at run
6345 time by the stack_trace_filter file in the debugfs
6346 tracing directory. Note, this enables stack tracing
6347 and the stacktrace above is not needed.
6351 Set the STI (builtin display/keyboard on the HP-PARISC
6352 machines) console (graphic card) which should be used
6353 as the initial boot-console.
6354 See also comment in drivers/video/console/sticore.c.
6357 See comment in drivers/video/console/sticore.c.
6360 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
6365 Enable or disable strict sigaltstack size checks
6366 against the required signal frame size which
6367 depends on the supported FPU features. This can
6368 be used to filter out binaries which have
6369 not yet been made aware of AT_MINSIGSTKSZ.
6371 stress_hpt [PPC,EARLY]
6372 Limits the number of kernel HPT entries in the hash
6373 page table to increase the rate of hash page table
6374 faults on kernel addresses.
6376 stress_slb [PPC,EARLY]
6377 Limits the number of kernel SLB entries, and flushes
6378 them frequently to increase the rate of SLB faults
6379 on kernel addresses.
6381 sunrpc.min_resvport=
6382 sunrpc.max_resvport=
6384 SunRPC servers often require that client requests
6385 originate from a privileged port (i.e. a port in the
6386 range 0 < portnr < 1024).
6387 An administrator who wishes to reserve some of these
6388 ports for other uses may adjust the range that the
6389 kernel's sunrpc client considers to be privileged
6390 using these two parameters to set the minimum and
6391 maximum port values.
6393 sunrpc.svc_rpc_per_connection_limit=
6395 Limit the number of requests that the server will
6396 process in parallel from a single connection.
6397 The default value is 0 (no limit).
6401 Control how the NFS server code allocates CPUs to
6402 service thread pools. Depending on how many NICs
6403 you have and where their interrupts are bound, this
6404 option will affect which CPUs will do NFS serving.
6405 Note: this parameter cannot be changed while the
6406 NFS server is running.
6408 auto the server chooses an appropriate mode
6409 automatically using heuristics
6410 global a single global pool contains all CPUs
6411 percpu one pool for each CPU
6412 pernode one pool for each NUMA node (equivalent
6413 to global on non-NUMA machines)
6415 sunrpc.tcp_slot_table_entries=
6416 sunrpc.udp_slot_table_entries=
6418 Sets the upper limit on the number of simultaneous
6419 RPC calls that can be sent from the client to a
6420 server. Increasing these values may allow you to
6421 improve throughput, but will also increase the
6422 amount of memory reserved for use by the client.
6424 suspend.pm_test_delay=
6426 Sets the number of seconds to remain in a suspend test
6427 mode before resuming the system (see
6428 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
6429 is set. Default value is 5.
6432 Format: { on | off | y | n | 1 | 0 }
6433 This parameter controls use of the Protected
6434 Execution Facility on pSeries.
6436 swiotlb= [ARM,IA-64,PPC,MIPS,X86,EARLY]
6437 Format: { <int> [,<int>] | force | noforce }
6438 <int> -- Number of I/O TLB slabs
6439 <int> -- Second integer after comma. Number of swiotlb
6440 areas with their own lock. Will be rounded up
6442 force -- force using of bounce buffers even if they
6443 wouldn't be automatically used by the kernel
6444 noforce -- Never use bounce buffers (for debugging)
6446 switches= [HW,M68k,EARLY]
6449 Set a sysctl parameter, right before loading the init
6450 process, as if the value was written to the respective
6451 /proc/sys/... file. Both '.' and '/' are recognized as
6452 separators. Unrecognized parameters and invalid values
6453 are reported in the kernel log. Sysctls registered
6454 later by a loaded module cannot be set this way.
6455 Example: sysctl.vm.swappiness=40
6457 sysrq_always_enabled
6459 Ignore sysrq setting - this boot parameter will
6460 neutralize any effect of /proc/sys/kernel/sysrq.
6461 Useful for debugging.
6463 tcpmhash_entries= [KNL,NET]
6464 Set the number of tcp_metrics_hash slots.
6465 Default value is 8192 or 16384 depending on total
6466 ram pages. This is used to specify the TCP metrics
6467 cache size. See Documentation/networking/ip-sysctl.rst
6468 "tcp_no_metrics_save" section for more details.
6472 test_suspend= [SUSPEND]
6473 Format: { "mem" | "standby" | "freeze" }[,N]
6474 Specify "mem" (for Suspend-to-RAM) or "standby" (for
6475 standby suspend) or "freeze" (for suspend type freeze)
6476 as the system sleep state during system startup with
6477 the optional capability to repeat N number of times.
6478 The system is woken from this state using a
6479 wakeup-capable RTC alarm.
6481 thash_entries= [KNL,NET]
6482 Set number of hash buckets for TCP connection
6484 thermal.act= [HW,ACPI]
6485 -1: disable all active trip points in all thermal zones
6486 <degrees C>: override all lowest active trip points
6488 thermal.crt= [HW,ACPI]
6489 -1: disable all critical trip points in all thermal zones
6490 <degrees C>: override all critical trip points
6492 thermal.off= [HW,ACPI]
6493 1: disable ACPI thermal control
6495 thermal.psv= [HW,ACPI]
6496 -1: disable all passive trip points
6497 <degrees C>: override all passive trip points to this
6500 thermal.tzp= [HW,ACPI]
6501 Specify global default ACPI thermal zone polling rate
6502 <deci-seconds>: poll all this frequency
6503 0: no polling (default)
6505 threadirqs [KNL,EARLY]
6506 Force threading of all interrupt handlers except those
6507 marked explicitly IRQF_NO_THREAD.
6509 topology= [S390,EARLY]
6511 Specify if the kernel should make use of the cpu
6512 topology information if the hardware supports this.
6513 The scheduler will make use of this information and
6514 e.g. base its process migration decisions on it.
6517 topology_updates= [KNL, PPC, NUMA]
6519 Specify if the kernel should ignore (off)
6520 topology updates sent by the hypervisor to this
6523 torture.disable_onoff_at_boot= [KNL]
6524 Prevent the CPU-hotplug component of torturing
6525 until after init has spawned.
6527 torture.ftrace_dump_at_shutdown= [KNL]
6528 Dump the ftrace buffer at torture-test shutdown,
6529 even if there were no errors. This can be a
6530 very costly operation when many torture tests
6531 are running concurrently, especially on systems
6532 with rotating-rust storage.
6534 torture.verbose_sleep_frequency= [KNL]
6535 Specifies how many verbose printk()s should be
6536 emitted between each sleep. The default of zero
6537 disables verbose-printk() sleeping.
6539 torture.verbose_sleep_duration= [KNL]
6540 Duration of each verbose-printk() sleep in jiffies.
6544 tpm_suspend_pcr=[HW,TPM]
6545 Format: integer pcr id
6546 Specify that at suspend time, the tpm driver
6547 should extend the specified pcr with zeros,
6548 as a workaround for some chips which fail to
6549 flush the last written pcr on TPM_SaveState.
6550 This will guarantee that all the other pcrs
6553 tpm_tis.interrupts= [HW,TPM]
6554 Enable interrupts for the MMIO based physical layer
6555 for the FIFO interface. By default it is set to false
6556 (0). For more information about TPM hardware interfaces
6557 defined by Trusted Computing Group (TCG) see
6558 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/
6561 Have the tracepoints sent to printk as well as the
6562 tracing ring buffer. This is useful for early boot up
6563 where the system hangs or reboots and does not give the
6564 option for reading the tracing buffer or performing a
6565 ftrace_dump_on_oops.
6567 To turn off having tracepoints sent to printk,
6568 echo 0 > /proc/sys/kernel/tracepoint_printk
6569 Note, echoing 1 into this file without the
6570 tracepoint_printk kernel cmdline option has no effect.
6572 The tp_printk_stop_on_boot (see below) can also be used
6573 to stop the printing of events to console at
6578 Having tracepoints sent to printk() and activating high
6579 frequency tracepoints such as irq or sched, can cause
6580 the system to live lock.
6582 tp_printk_stop_on_boot [FTRACE]
6583 When tp_printk (above) is set, it can cause a lot of noise
6584 on the console. It may be useful to only include the
6585 printing of events during boot up, as user space may
6586 make the system inoperable.
6588 This command line option will stop the printing of events
6589 to console at the late_initcall_sync() time frame.
6591 trace_buf_size=nn[KMG]
6592 [FTRACE] will set tracing buffer size on each cpu.
6594 trace_clock= [FTRACE] Set the clock used for tracing events
6596 local - Use the per CPU time stamp counter
6597 (converted into nanoseconds). Fast, but
6598 depending on the architecture, may not be
6599 in sync between CPUs.
6600 global - Event time stamps are synchronize across
6601 CPUs. May be slower than the local clock,
6602 but better for some race conditions.
6603 counter - Simple counting of events (1, 2, ..)
6604 note, some counts may be skipped due to the
6605 infrastructure grabbing the clock more than
6607 uptime - Use jiffies as the time stamp.
6608 perf - Use the same clock that perf uses.
6609 mono - Use ktime_get_mono_fast_ns() for time stamps.
6610 mono_raw - Use ktime_get_raw_fast_ns() for time
6612 boot - Use ktime_get_boot_fast_ns() for time stamps.
6613 Architectures may add more clocks. See
6614 Documentation/trace/ftrace.rst for more details.
6616 trace_event=[event-list]
6617 [FTRACE] Set and start specified trace events in order
6618 to facilitate early boot debugging. The event-list is a
6619 comma-separated list of trace events to enable. See
6620 also Documentation/trace/events.rst
6622 trace_instance=[instance-info]
6623 [FTRACE] Create a ring buffer instance early in boot up.
6624 This will be listed in:
6626 /sys/kernel/tracing/instances
6628 Events can be enabled at the time the instance is created
6631 trace_instance=<name>,<system1>:<event1>,<system2>:<event2>
6633 Note, the "<system*>:" portion is optional if the event is
6636 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall
6638 will enable the "sched_switch" event (note, the "sched:" is optional, and
6639 the same thing would happen if it was left off). The irq_handler_entry
6640 event, and all events under the "initcall" system.
6642 trace_options=[option-list]
6643 [FTRACE] Enable or disable tracer options at boot.
6644 The option-list is a comma delimited list of options
6645 that can be enabled or disabled just as if you were
6646 to echo the option name into
6648 /sys/kernel/tracing/trace_options
6650 For example, to enable stacktrace option (to dump the
6651 stack trace of each event), add to the command line:
6653 trace_options=stacktrace
6655 See also Documentation/trace/ftrace.rst "trace options"
6658 trace_trigger=[trigger-list]
6659 [FTRACE] Add a event trigger on specific events.
6660 Set a trigger on top of a specific event, with an optional
6663 The format is is "trace_trigger=<event>.<trigger>[ if <filter>],..."
6664 Where more than one trigger may be specified that are comma deliminated.
6668 trace_trigger="sched_switch.stacktrace if prev_state == 2"
6670 The above will enable the "stacktrace" trigger on the "sched_switch"
6671 event but only trigger it if the "prev_state" of the "sched_switch"
6672 event is "2" (TASK_UNINTERUPTIBLE).
6674 See also "Event triggers" in Documentation/trace/events.rst
6678 [FTRACE] enable this option to disable tracing when a
6679 warning is hit. This turns off "tracing_on". Tracing can
6680 be enabled again by echoing '1' into the "tracing_on"
6681 file located in /sys/kernel/tracing/
6683 This option is useful, as it disables the trace before
6684 the WARNING dump is called, which prevents the trace to
6685 be filled with content caused by the warning output.
6687 This option can also be set at run time via the sysctl
6688 option: kernel/traceoff_on_warning
6690 transparent_hugepage=
6692 Format: [always|madvise|never]
6693 Can be used to control the default behavior of the system
6694 with respect to transparent hugepages.
6695 See Documentation/admin-guide/mm/transhuge.rst
6698 trusted.source= [KEYS]
6700 This parameter identifies the trust source as a backend
6701 for trusted keys implementation. Supported trust
6706 If not specified then it defaults to iterating through
6707 the trust source list starting with TPM and assigns the
6708 first trust source as a backend which is initialized
6709 successfully during iteration.
6713 The RNG used to generate key material for trusted keys.
6716 - the same value as trusted.source: "tpm" or "tee"
6718 If not specified, "default" is used. In this case,
6719 the RNG's choice is left to each individual trust source.
6721 tsc= Disable clocksource stability checks for TSC.
6723 [x86] reliable: mark tsc clocksource as reliable, this
6724 disables clocksource verification at runtime, as well
6725 as the stability checks done at bootup. Used to enable
6726 high-resolution timer mode on older hardware, and in
6727 virtualized environment.
6728 [x86] noirqtime: Do not use TSC to do irq accounting.
6729 Used to run time disable IRQ_TIME_ACCOUNTING on any
6730 platforms where RDTSC is slow and this accounting
6732 [x86] unstable: mark the TSC clocksource as unstable, this
6733 marks the TSC unconditionally unstable at bootup and
6734 avoids any further wobbles once the TSC watchdog notices.
6735 [x86] nowatchdog: disable clocksource watchdog. Used
6736 in situations with strict latency requirements (where
6737 interruptions from clocksource watchdog are not
6739 [x86] recalibrate: force recalibration against a HW timer
6740 (HPET or PM timer) on systems whose TSC frequency was
6741 obtained from HW or FW using either an MSR or CPUID(0x15).
6742 Warn if the difference is more than 500 ppm.
6743 [x86] watchdog: Use TSC as the watchdog clocksource with
6744 which to check other HW timers (HPET or PM timer), but
6745 only on systems where TSC has been deemed trustworthy.
6746 This will be suppressed by an earlier tsc=nowatchdog and
6747 can be overridden by a later tsc=nowatchdog. A console
6748 message will flag any such suppression or overriding.
6750 tsc_early_khz= [X86,EARLY] Skip early TSC calibration and use the given
6751 value instead. Useful when the early TSC frequency discovery
6752 procedure is not reliable, such as on overclocked systems
6753 with CPUID.16h support and partial CPUID.15h support.
6754 Format: <unsigned int>
6756 tsx= [X86] Control Transactional Synchronization
6757 Extensions (TSX) feature in Intel processors that
6758 support TSX control.
6760 This parameter controls the TSX feature. The options are:
6762 on - Enable TSX on the system. Although there are
6763 mitigations for all known security vulnerabilities,
6764 TSX has been known to be an accelerator for
6765 several previous speculation-related CVEs, and
6766 so there may be unknown security risks associated
6767 with leaving it enabled.
6769 off - Disable TSX on the system. (Note that this
6770 option takes effect only on newer CPUs which are
6771 not vulnerable to MDS, i.e., have
6772 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
6773 the new IA32_TSX_CTRL MSR through a microcode
6774 update. This new MSR allows for the reliable
6775 deactivation of the TSX functionality.)
6777 auto - Disable TSX if X86_BUG_TAA is present,
6778 otherwise enable TSX on the system.
6780 Not specifying this option is equivalent to tsx=off.
6782 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6785 tsx_async_abort= [X86,INTEL,EARLY] Control mitigation for the TSX Async
6786 Abort (TAA) vulnerability.
6788 Similar to Micro-architectural Data Sampling (MDS)
6789 certain CPUs that support Transactional
6790 Synchronization Extensions (TSX) are vulnerable to an
6791 exploit against CPU internal buffers which can forward
6792 information to a disclosure gadget under certain
6795 In vulnerable processors, the speculatively forwarded
6796 data can be used in a cache side channel attack, to
6797 access data to which the attacker does not have direct
6800 This parameter controls the TAA mitigation. The
6803 full - Enable TAA mitigation on vulnerable CPUs
6806 full,nosmt - Enable TAA mitigation and disable SMT on
6807 vulnerable CPUs. If TSX is disabled, SMT
6808 is not disabled because CPU is not
6809 vulnerable to cross-thread TAA attacks.
6810 off - Unconditionally disable TAA mitigation
6812 On MDS-affected machines, tsx_async_abort=off can be
6813 prevented by an active MDS mitigation as both vulnerabilities
6814 are mitigated with the same mechanism so in order to disable
6815 this mitigation, you need to specify mds=off too.
6817 Not specifying this option is equivalent to
6818 tsx_async_abort=full. On CPUs which are MDS affected
6819 and deploy MDS mitigation, TAA mitigation is not
6820 required and doesn't provide any additional
6824 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6826 turbografx.map[2|3]= [HW,JOY]
6827 TurboGraFX parallel port interface
6829 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
6830 See also Documentation/input/devices/joystick-parport.rst
6832 udbg-immortal [PPC] When debugging early kernel crashes that
6833 happen after console_init() and before a proper
6834 console driver takes over, this boot options might
6835 help "seeing" what's going on.
6837 uhash_entries= [KNL,NET]
6838 Set number of hash buckets for UDP/UDP-Lite connections
6841 [USB] Ignore overcurrent events (default N).
6842 Some badly-designed motherboards generate lots of
6843 bogus events, for ports that aren't wired to
6844 anything. Set this parameter to avoid log spamming.
6845 Note that genuine overcurrent events won't be
6849 [X86] Cause panic on unknown NMI.
6851 unwind_debug [X86-64,EARLY]
6852 Enable unwinder debug output. This can be
6853 useful for debugging certain unwinder error
6854 conditions, including corrupt stacks and
6855 bad/missing unwinder metadata.
6857 usbcore.authorized_default=
6858 [USB] Default USB device authorization:
6859 (default -1 = authorized (same as 1),
6860 0 = not authorized, 1 = authorized, 2 = authorized
6861 if device connected to internal port)
6863 usbcore.autosuspend=
6864 [USB] The autosuspend time delay (in seconds) used
6865 for newly-detected USB devices (default 2). This
6866 is the time required before an idle device will be
6867 autosuspended. Devices for which the delay is set
6868 to a negative value won't be autosuspended at all.
6870 usbcore.usbfs_snoop=
6871 [USB] Set to log all usbfs traffic (default 0 = off).
6873 usbcore.usbfs_snoop_max=
6874 [USB] Maximum number of bytes to snoop in each URB
6877 usbcore.blinkenlights=
6878 [USB] Set to cycle leds on hubs (default 0 = off).
6880 usbcore.old_scheme_first=
6881 [USB] Start with the old device initialization
6882 scheme (default 0 = off).
6884 usbcore.usbfs_memory_mb=
6885 [USB] Memory limit (in MB) for buffers allocated by
6886 usbfs (default = 16, 0 = max = 2047).
6888 usbcore.use_both_schemes=
6889 [USB] Try the other device initialization scheme
6890 if the first one fails (default 1 = enabled).
6892 usbcore.initial_descriptor_timeout=
6893 [USB] Specifies timeout for the initial 64-byte
6894 USB_REQ_GET_DESCRIPTOR request in milliseconds
6895 (default 5000 = 5.0 seconds).
6897 usbcore.nousb [USB] Disable the USB subsystem
6900 [USB] A list of quirk entries to augment the built-in
6901 usb core quirk list. List entries are separated by
6902 commas. Each entry has the form
6903 VendorID:ProductID:Flags. The IDs are 4-digit hex
6904 numbers and Flags is a set of letters. Each letter
6905 will change the built-in quirk; setting it if it is
6906 clear and clearing it if it is set. The letters have
6907 the following meanings:
6908 a = USB_QUIRK_STRING_FETCH_255 (string
6909 descriptors must not be fetched using
6911 b = USB_QUIRK_RESET_RESUME (device can't resume
6912 correctly so reset it instead);
6913 c = USB_QUIRK_NO_SET_INTF (device can't handle
6914 Set-Interface requests);
6915 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
6916 handle its Configuration or Interface
6918 e = USB_QUIRK_RESET (device can't be reset
6919 (e.g morph devices), don't use reset);
6920 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
6921 more interface descriptions than the
6922 bNumInterfaces count, and can't handle
6923 talking to these interfaces);
6924 g = USB_QUIRK_DELAY_INIT (device needs a pause
6925 during initialization, after we read
6926 the device descriptor);
6927 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
6928 high speed and super speed interrupt
6929 endpoints, the USB 2.0 and USB 3.0 spec
6930 require the interval in microframes (1
6931 microframe = 125 microseconds) to be
6932 calculated as interval = 2 ^
6934 Devices with this quirk report their
6935 bInterval as the result of this
6936 calculation instead of the exponent
6937 variable used in the calculation);
6938 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
6939 handle device_qualifier descriptor
6941 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
6942 generates spurious wakeup, ignore
6943 remote wakeup capability);
6944 k = USB_QUIRK_NO_LPM (device can't handle Link
6946 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
6947 (Device reports its bInterval as linear
6948 frames instead of the USB 2.0
6950 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
6951 to be disconnected before suspend to
6952 prevent spurious wakeup);
6953 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
6954 pause after every control message);
6955 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
6956 delay after resetting its port);
6957 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT
6958 (Reduce timeout of the SET_ADDRESS
6959 request from 5000 ms to 500 ms);
6960 Example: quirks=0781:5580:bk,0a5c:5834:gij
6963 [USBHID] The interval which mice are to be polled at.
6966 [USBHID] The interval which joysticks are to be polled at.
6969 [USBHID] The interval which keyboards are to be polled at.
6971 usb-storage.delay_use=
6972 [UMS] The delay in seconds before a new device is
6973 scanned for Logical Units (default 1).
6976 [UMS] A list of quirks entries to supplement or
6977 override the built-in unusual_devs list. List
6978 entries are separated by commas. Each entry has
6979 the form VID:PID:Flags where VID and PID are Vendor
6980 and Product ID values (4-digit hex numbers) and
6981 Flags is a set of characters, each corresponding
6982 to a common usb-storage quirk flag as follows:
6983 a = SANE_SENSE (collect more than 18 bytes
6984 of sense data, not on uas);
6985 b = BAD_SENSE (don't collect more than 18
6986 bytes of sense data, not on uas);
6987 c = FIX_CAPACITY (decrease the reported
6988 device capacity by one sector);
6989 d = NO_READ_DISC_INFO (don't use
6990 READ_DISC_INFO command, not on uas);
6991 e = NO_READ_CAPACITY_16 (don't use
6992 READ_CAPACITY_16 command);
6993 f = NO_REPORT_OPCODES (don't use report opcodes
6995 g = MAX_SECTORS_240 (don't transfer more than
6996 240 sectors at a time, uas only);
6997 h = CAPACITY_HEURISTICS (decrease the
6998 reported device capacity by one
6999 sector if the number is odd);
7000 i = IGNORE_DEVICE (don't bind to this
7002 j = NO_REPORT_LUNS (don't use report luns
7004 k = NO_SAME (do not use WRITE_SAME, uas only)
7005 l = NOT_LOCKABLE (don't try to lock and
7006 unlock ejectable media, not on uas);
7007 m = MAX_SECTORS_64 (don't transfer more
7008 than 64 sectors = 32 KB at a time,
7010 n = INITIAL_READ10 (force a retry of the
7011 initial READ(10) command, not on uas);
7012 o = CAPACITY_OK (accept the capacity
7013 reported by the device, not on uas);
7014 p = WRITE_CACHE (the device cache is ON
7015 by default, not on uas);
7016 r = IGNORE_RESIDUE (the device reports
7017 bogus residue values, not on uas);
7018 s = SINGLE_LUN (the device has only one
7020 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
7021 commands, uas only);
7022 u = IGNORE_UAS (don't bind to the uas driver);
7023 w = NO_WP_DETECT (don't test whether the
7024 medium is write-protected).
7025 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
7026 even if the device claims no cache,
7028 Example: quirks=0419:aaf5:rl,0421:0433:rc
7030 user_debug= [KNL,ARM]
7032 See arch/arm/Kconfig.debug help text.
7033 1 - undefined instruction events
7035 4 - invalid data aborts
7038 Example: user_debug=31
7041 [X86,EARLY] Flags controlling user PTE allocations.
7043 nohigh = do not allocate PTE pages in
7044 HIGHMEM regardless of setting
7047 vdso= [X86,SH,SPARC]
7048 On X86_32, this is an alias for vdso32=. Otherwise:
7050 vdso=1: enable VDSO (the default)
7051 vdso=0: disable VDSO mapping
7053 vdso32= [X86] Control the 32-bit vDSO
7054 vdso32=1: enable 32-bit VDSO
7055 vdso32=0 or vdso32=2: disable 32-bit VDSO
7057 See the help text for CONFIG_COMPAT_VDSO for more
7058 details. If CONFIG_COMPAT_VDSO is set, the default is
7059 vdso32=0; otherwise, the default is vdso32=1.
7061 For compatibility with older kernels, vdso32=2 is an
7064 Try vdso32=0 if you encounter an error that says:
7065 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
7068 vector=percpu: enable percpu vector domain
7070 video= [FB,EARLY] Frame buffer configuration
7071 See Documentation/fb/modedb.rst.
7073 video.brightness_switch_enabled= [ACPI]
7075 If set to 1, on receiving an ACPI notify event
7076 generated by hotkey, video driver will adjust brightness
7077 level and then send out the event to user space through
7078 the allocated input device. If set to 0, video driver
7079 will only send out the event without touching backlight
7084 [VMMIO] Memory mapped virtio (platform) device.
7086 <size>@<baseaddr>:<irq>[:<id>]
7088 <size> := size (can use standard suffixes
7090 <baseaddr> := physical base address
7091 <irq> := interrupt number (as passed to
7093 <id> := (optional) platform device id
7095 virtio_mmio.device=1K@0x100b0000:48:7
7097 Can be used multiple times for multiple devices.
7099 vga= [BOOT,X86-32] Select a particular video mode
7100 See Documentation/arch/x86/boot.rst and
7101 Documentation/admin-guide/svga.rst.
7102 Use vga=ask for menu.
7103 This is actually a boot loader parameter; the value is
7104 passed to the kernel using a special protocol.
7106 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
7107 May slow down system boot speed, especially when
7108 enabled on systems with a large amount of memory.
7109 All options are enabled by default, and this
7110 interface is meant to allow for selectively
7111 enabling or disabling specific virtual memory
7114 Available options are:
7115 P Enable page structure init time poisoning
7116 - Disable all of the above options
7118 vmalloc=nn[KMG] [KNL,BOOT,EARLY] Forces the vmalloc area to have an
7119 exact size of <nn>. This can be used to increase
7120 the minimum size (128MB on x86). It can also be
7121 used to decrease the size and leave more room
7122 for directly mapped kernel RAM.
7124 vmcp_cma=nn[MG] [KNL,S390,EARLY]
7125 Sets the memory size reserved for contiguous memory
7126 allocations for the vmcp device driver.
7128 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
7131 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
7134 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
7137 vsyscall= [X86-64,EARLY]
7138 Controls the behavior of vsyscalls (i.e. calls to
7139 fixed addresses of 0xffffffffff600x00 from legacy
7140 code). Most statically-linked binaries and older
7141 versions of glibc use these calls. Because these
7142 functions are at fixed addresses, they make nice
7143 targets for exploits that can control RIP.
7145 emulate Vsyscalls turn into traps and are emulated
7146 reasonably safely. The vsyscall page is
7149 xonly [default] Vsyscalls turn into traps and are
7150 emulated reasonably safely. The vsyscall
7151 page is not readable.
7153 none Vsyscalls don't work at all. This makes
7154 them quite hard to use for exploits but
7155 might break your system.
7157 vt.color= [VT] Default text color.
7158 Format: 0xYX, X = foreground, Y = background.
7159 Default: 0x07 = light gray on black.
7161 vt.cur_default= [VT] Default cursor shape.
7162 Format: 0xCCBBAA, where AA, BB, and CC are the same as
7163 the parameters of the <Esc>[?A;B;Cc escape sequence;
7164 see VGA-softcursor.txt. Default: 2 = underline.
7166 vt.default_blu= [VT]
7167 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
7168 Change the default blue palette of the console.
7169 This is a 16-member array composed of values
7172 vt.default_grn= [VT]
7173 Format: <green0>,<green1>,<green2>,...,<green15>
7174 Change the default green palette of the console.
7175 This is a 16-member array composed of values
7178 vt.default_red= [VT]
7179 Format: <red0>,<red1>,<red2>,...,<red15>
7180 Change the default red palette of the console.
7181 This is a 16-member array composed of values
7187 Set system-wide default UTF-8 mode for all tty's.
7188 Default is 1, i.e. UTF-8 mode is enabled for all
7189 newly opened terminals.
7191 vt.global_cursor_default=
7194 Set system-wide default for whether a cursor
7195 is shown on new VTs. Default is -1,
7196 i.e. cursors will be created by default unless
7197 overridden by individual drivers. 0 will hide
7198 cursors, 1 will display them.
7200 vt.italic= [VT] Default color for italic text; 0-15.
7203 vt.underline= [VT] Default color for underlined text; 0-15.
7206 watchdog timers [HW,WDT] For information on watchdog timers,
7207 see Documentation/watchdog/watchdog-parameters.rst
7208 or other driver-specific files in the
7209 Documentation/watchdog/ directory.
7213 Set the hard lockup detector stall duration
7214 threshold in seconds. The soft lockup detector
7215 threshold is set to twice the value. A value of 0
7216 disables both lockup detectors. Default is 10
7219 workqueue.unbound_cpus=
7220 [KNL,SMP] Specify to constrain one or some CPUs
7221 to use in unbound workqueues.
7223 By default, all online CPUs are available for
7226 workqueue.watchdog_thresh=
7227 If CONFIG_WQ_WATCHDOG is configured, workqueue can
7228 warn stall conditions and dump internal state to
7229 help debugging. 0 disables workqueue stall
7230 detection; otherwise, it's the stall threshold
7231 duration in seconds. The default value is 30 and
7232 it can be updated at runtime by writing to the
7233 corresponding sysfs file.
7235 workqueue.cpu_intensive_thresh_us=
7236 Per-cpu work items which run for longer than this
7237 threshold are automatically considered CPU intensive
7238 and excluded from concurrency management to prevent
7239 them from noticeably delaying other per-cpu work
7240 items. Default is 10000 (10ms).
7242 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel
7243 will report the work functions which violate this
7244 threshold repeatedly. They are likely good
7245 candidates for using WQ_UNBOUND workqueues instead.
7247 workqueue.power_efficient
7248 Per-cpu workqueues are generally preferred because
7249 they show better performance thanks to cache
7250 locality; unfortunately, per-cpu workqueues tend to
7251 be more power hungry than unbound workqueues.
7253 Enabling this makes the per-cpu workqueues which
7254 were observed to contribute significantly to power
7255 consumption unbound, leading to measurably lower
7256 power usage at the cost of small performance
7259 The default value of this parameter is determined by
7260 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
7262 workqueue.default_affinity_scope=
7263 Select the default affinity scope to use for unbound
7264 workqueues. Can be one of "cpu", "smt", "cache",
7265 "numa" and "system". Default is "cache". For more
7266 information, see the Affinity Scopes section in
7267 Documentation/core-api/workqueue.rst.
7269 This can be changed after boot by writing to the
7270 matching /sys/module/workqueue/parameters file. All
7271 workqueues with the "default" affinity scope will be
7272 updated accordignly.
7274 workqueue.debug_force_rr_cpu
7275 Workqueue used to implicitly guarantee that work
7276 items queued without explicit CPU specified are put
7277 on the local CPU. This guarantee is no longer true
7278 and while local CPU is still preferred work items
7279 may be put on foreign CPUs. This debug option
7280 forces round-robin CPU selection to flush out
7281 usages which depend on the now broken guarantee.
7282 When enabled, memory and cache locality will be
7285 writecombine= [LOONGARCH,EARLY] Control the MAT (Memory Access
7286 Type) of ioremap_wc().
7288 on - Enable writecombine, use WUC for ioremap_wc()
7289 off - Disable writecombine, use SUC for ioremap_wc()
7291 x2apic_phys [X86-64,APIC,EARLY] Use x2apic physical mode instead of
7292 default x2apic cluster mode on platforms
7295 xen_512gb_limit [KNL,X86-64,XEN]
7296 Restricts the kernel running paravirtualized under Xen
7297 to use only up to 512 GB of RAM. The reason to do so is
7298 crash analysis tools and Xen tools for doing domain
7299 save/restore/migration must be enabled to handle larger
7302 xen_emul_unplug= [HW,X86,XEN,EARLY]
7303 Unplug Xen emulated devices
7304 Format: [unplug0,][unplug1]
7305 ide-disks -- unplug primary master IDE devices
7306 aux-ide-disks -- unplug non-primary-master IDE devices
7307 nics -- unplug network devices
7308 all -- unplug all emulated devices (NICs and IDE disks)
7309 unnecessary -- unplugging emulated devices is
7310 unnecessary even if the host did not respond to
7312 never -- do not unplug even if version check succeeds
7314 xen_legacy_crash [X86,XEN,EARLY]
7315 Crash from Xen panic notifier, without executing late
7316 panic() code such as dumping handler.
7318 xen_msr_safe= [X86,XEN,EARLY]
7320 Select whether to always use non-faulting (safe) MSR
7321 access functions when running as Xen PV guest. The
7322 default value is controlled by CONFIG_XEN_PV_MSR_SAFE.
7324 xen_nopvspin [X86,XEN,EARLY]
7325 Disables the qspinlock slowpath using Xen PV optimizations.
7326 This parameter is obsoleted by "nopvspin" parameter, which
7327 has equivalent effect for XEN platform.
7330 Disables the PV optimizations forcing the HVM guest to
7331 run as generic HVM guest with no PV drivers.
7332 This option is obsoleted by the "nopv" option, which
7333 has equivalent effect for XEN platform.
7335 xen_no_vector_callback
7336 [KNL,X86,XEN,EARLY] Disable the vector callback for Xen
7337 event channel interrupts.
7339 xen_scrub_pages= [XEN]
7340 Boolean option to control scrubbing pages before giving them back
7341 to Xen, for use by other domains. Can be also changed at runtime
7342 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
7343 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
7345 xen_timer_slop= [X86-64,XEN,EARLY]
7346 Set the timer slop (in nanoseconds) for the virtual Xen
7347 timers (default is 100000). This adjusts the minimum
7348 delta of virtualized Xen timers, where lower values
7349 improve timer resolution at the expense of processing
7350 more timer interrupts.
7352 xen.balloon_boot_timeout= [XEN]
7353 The time (in seconds) to wait before giving up to boot
7354 in case initial ballooning fails to free enough memory.
7355 Applies only when running as HVM or PVH guest and
7356 started with less memory configured than allowed at
7357 max. Default is 180.
7359 xen.event_eoi_delay= [XEN]
7360 How long to delay EOI handling in case of event
7361 storms (jiffies). Default is 10.
7363 xen.event_loop_timeout= [XEN]
7364 After which time (jiffies) the event handling loop
7365 should start to delay EOI handling. Default is 2.
7367 xen.fifo_events= [XEN]
7368 Boolean parameter to disable using fifo event handling
7369 even if available. Normally fifo event handling is
7370 preferred over the 2-level event handling, as it is
7371 fairer and the number of possible event channels is
7372 much higher. Default is on (use fifo events).
7374 xirc2ps_cs= [NET,PCMCIA]
7376 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
7379 By default on POWER9 and above, the kernel will
7380 natively use the XIVE interrupt controller. This option
7381 allows the fallback firmware mode to be used:
7383 off Fallback to firmware control of XIVE interrupt
7384 controller on both pseries and powernv
7385 platforms. Only useful on POWER9 and above.
7387 xive.store-eoi=off [PPC]
7388 By default on POWER10 and above, the kernel will use
7389 stores for EOI handling when the XIVE interrupt mode
7390 is active. This option allows the XIVE driver to use
7391 loads instead, as on POWER9.
7393 xhci-hcd.quirks [USB,KNL]
7394 A hex value specifying bitmask with supplemental xhci
7395 host controller quirks. Meaning of each bit can be
7396 consulted in header drivers/usb/host/xhci.h.
7399 Format: { early | on | rw | ro | off }
7400 Controls if xmon debugger is enabled. Default is off.
7401 Passing only "xmon" is equivalent to "xmon=early".
7402 early Call xmon as early as possible on boot; xmon
7403 debugger is called from setup_arch().
7404 on xmon debugger hooks will be installed so xmon
7405 is only called on a kernel crash. Default mode,
7406 i.e. either "ro" or "rw" mode, is controlled
7407 with CONFIG_XMON_DEFAULT_RO_MODE.
7408 rw xmon debugger hooks will be installed so xmon
7409 is called only on a kernel crash, mode is write,
7410 meaning SPR registers, memory and, other data
7411 can be written using xmon commands.
7412 ro same as "rw" option above but SPR registers,
7413 memory, and other data can't be written using
7415 off xmon is disabled.
projects / linux-2.6-microblaze.git / blob