1 AMD64 specific boot options
3 There are many others (usually documented in driver documentation), but
4 only the AMD64 specific ones are listed here.
8 Please see Documentation/x86/x86_64/machinecheck for sysfs runtime tunables.
13 Disable CMCI(Corrected Machine Check Interrupt) that
14 Intel processor supports. Usually this disablement is
15 not recommended, but it might be handy if your hardware
17 Note that you'll get more problems without CMCI than with
18 due to the shared banks, i.e. you might get duplicated
21 Don't make logs for corrected errors. All events reported
22 as corrected are silently cleared by OS.
23 This option will be useful if you have no interest in any
26 Disable features for corrected errors, e.g. polling timer
27 and CMCI. All events reported as corrected are not cleared
28 by OS and remained in its error banks.
29 Usually this disablement is not recommended, however if
30 there is an agent checking/clearing corrected errors
31 (e.g. BIOS or hardware monitoring applications), conflicting
32 with OS's error handling, and you cannot deactivate the agent,
33 then this option will be a help.
35 Enable logging of machine checks left over from booting.
36 Disabled by default on AMD because some BIOS leave bogus ones.
37 If your BIOS doesn't do that it's a good idea to enable though
38 to make sure you log even machine check events that result
39 in a reboot. On Intel systems it is enabled by default.
41 Disable boot machine check logging.
42 mce=tolerancelevel[,monarchtimeout] (number,number)
44 0: always panic on uncorrected errors, log corrected errors
45 1: panic or SIGBUS on uncorrected errors, log corrected errors
46 2: SIGBUS or log uncorrected errors, log corrected errors
47 3: never panic or SIGBUS, log all errors (for testing only)
49 Can be also set using sysfs which is preferable.
51 Sets the time in us to wait for other CPUs on machine checks. 0
54 nomce (for compatibility with i386): same as mce=off
56 Everything else is in sysfs now.
60 apic Use IO-APIC. Default
62 noapic Don't use the IO-APIC.
64 disableapic Don't use the local APIC
66 nolapic Don't use the local APIC (alias for i386 compatibility)
68 pirq=... See Documentation/x86/i386/IO-APIC.txt
70 noapictimer Don't set up the APIC timer
72 no_timer_check Don't check the IO-APIC timer. This can work around
73 problems with incorrect timer initialization on some boards.
75 apicmaintimer Run time keeping from the local APIC timer instead
76 of using the PIT/HPET interrupt for this. This is useful
77 when the PIT/HPET interrupts are unreliable.
79 noapicmaintimer Don't do time keeping using the APIC timer.
80 Useful when this option was auto selected, but doesn't work.
83 Do APIC timer calibration using the pmtimer. Implies
84 apicmaintimer. Useful when your PIT timer is totally
89 syntax: earlyprintk=vga
90 earlyprintk=serial[,ttySn[,baudrate]]
92 The early console is useful when the kernel crashes before the
93 normal console is initialized. It is not enabled by
94 default because it has some cosmetic problems.
95 Append ,keep to not disable it when the real console takes over.
96 Only vga or serial at a time, not both.
97 Currently only ttyS0 and ttyS1 are supported.
98 Interaction with the standard serial driver is not very good.
99 The VGA output is eventually overwritten by the real console.
104 Don't use the CPU time stamp counter to read the wall time.
105 This can be used to work around timing problems on multiprocessor systems
106 with not properly synchronized CPUs.
109 Report when timer interrupts are lost because some code turned off
110 interrupts for too long.
113 Don't use the HPET timer.
118 Don't do power saving in the idle loop using HLT, but poll for rescheduling
119 event. This will make the CPUs eat a lot more power, but may be useful
120 to get slightly better performance in multiprocessor benchmarks. It also
121 makes some profiling using performance counters more accurate.
122 Please note that on systems with MONITOR/MWAIT support (like Intel EM64T
123 CPUs) this option has no performance advantage over the normal idle loop.
124 It may also interact badly with hyperthreading.
128 reboot=b[ios] | t[riple] | k[bd] | a[cpi] | e[fi] [, [w]arm | [c]old]
129 bios Use the CPU reboot vector for warm reset
130 warm Don't set the cold reboot flag
131 cold Set the cold reboot flag
132 triple Force a triple fault (init)
133 kbd Use the keyboard controller. cold reset (default)
134 acpi Use the ACPI RESET_REG in the FADT. If ACPI is not configured or the
135 ACPI reset does not work, the reboot path attempts the reset using
136 the keyboard controller.
137 efi Use efi reset_system runtime service. If EFI is not configured or the
138 EFI reset does not work, the reboot path attempts the reset using
139 the keyboard controller.
141 Using warm reset will be much faster especially on big memory
142 systems because the BIOS will not go through the memory check.
143 Disadvantage is that not all hardware will be completely reinitialized
144 on reboot so there may be boot problems on some systems.
148 Don't stop other CPUs on reboot. This can make reboot more reliable
151 Non Executable Mappings
160 additional_cpus=NUM Allow NUM more CPUs for hotplug
161 (defaults are specified by the BIOS, see Documentation/x86/x86_64/cpu-hotplug-spec)
165 numa=off Only set up a single NUMA node spanning all memory.
167 numa=noacpi Don't parse the SRAT table for NUMA setup
170 If given as a memory unit, fills all system RAM with nodes of
171 size interleaved over physical nodes.
174 If a number, fakes CMDLINE nodes and ignores NUMA setup of the
175 actual machine. Otherwise, system memory is configured
176 depending on the sizes and coefficients listed. For example:
177 numa=fake=2*512,1024,4*256,*128
178 gives two 512M nodes, a 1024M node, four 256M nodes, and the
179 rest split into 128M chunks. If the last character of CMDLINE
180 is a *, the remaining memory is divided up equally among its
183 gives two 512M nodes and the rest split into two nodes.
184 Otherwise, the remaining system RAM is allocated to an
189 acpi=off Don't enable ACPI
190 acpi=ht Use ACPI boot table parsing, but don't enable ACPI
192 acpi=force Force ACPI on (currently not needed)
194 acpi=strict Disable out of spec ACPI workarounds.
196 acpi_sci={edge,level,high,low} Set up ACPI SCI interrupt.
198 acpi=noirq Don't route interrupts
202 pci=off Don't use PCI
203 pci=conf1 Use conf1 access.
204 pci=conf2 Use conf2 access.
206 pci=assign-busses Assign busses
207 pci=irqmask=MASK Set PCI interrupt mask to MASK
208 pci=lastbus=NUMBER Scan upto NUMBER busses, no matter what the mptable says.
209 pci=noacpi Don't use ACPI to set up PCI interrupt routing.
211 IOMMU (input/output memory management unit)
213 Currently four x86-64 PCI-DMA mapping implementations exist:
215 1. <arch/x86_64/kernel/pci-nommu.c>: use no hardware/software IOMMU at all
216 (e.g. because you have < 3 GB memory).
217 Kernel boot message: "PCI-DMA: Disabling IOMMU"
219 2. <arch/x86_64/kernel/pci-gart.c>: AMD GART based hardware IOMMU.
220 Kernel boot message: "PCI-DMA: using GART IOMMU"
222 3. <arch/x86_64/kernel/pci-swiotlb.c> : Software IOMMU implementation. Used
223 e.g. if there is no hardware IOMMU in the system and it is need because
224 you have >3GB memory or told the kernel to us it (iommu=soft))
225 Kernel boot message: "PCI-DMA: Using software bounce buffering
228 4. <arch/x86_64/pci-calgary.c> : IBM Calgary hardware IOMMU. Used in IBM
229 pSeries and xSeries servers. This hardware IOMMU supports DMA address
230 mapping with memory protection, etc.
231 Kernel boot message: "PCI-DMA: Using Calgary IOMMU"
233 iommu=[<size>][,noagp][,off][,force][,noforce][,leak[=<nr_of_leak_pages>]
234 [,memaper[=<order>]][,merge][,forcesac][,fullflush][,nomerge]
235 [,noaperture][,calgary]
237 General iommu options:
238 off Don't initialize and use any kind of IOMMU.
239 noforce Don't force hardware IOMMU usage when it is not needed.
241 force Force the use of the hardware IOMMU even when it is
242 not actually needed (e.g. because < 3 GB memory).
243 soft Use software bounce buffering (SWIOTLB) (default for
244 Intel machines). This can be used to prevent the usage
245 of an available hardware IOMMU.
247 iommu options only relevant to the AMD GART hardware IOMMU:
248 <size> Set the size of the remapping area in bytes.
249 allowed Overwrite iommu off workarounds for specific chipsets.
250 fullflush Flush IOMMU on each allocation (default).
251 nofullflush Don't use IOMMU fullflush.
252 leak Turn on simple iommu leak tracing (only when
253 CONFIG_IOMMU_LEAK is on). Default number of leak pages
255 memaper[=<order>] Allocate an own aperture over RAM with size 32MB<<order.
256 (default: order=1, i.e. 64MB)
257 merge Do scatter-gather (SG) merging. Implies "force"
259 nomerge Don't do scatter-gather (SG) merging.
260 noaperture Ask the IOMMU not to touch the aperture for AGP.
261 forcesac Force single-address cycle (SAC) mode for masks <40bits
263 noagp Don't initialize the AGP driver and use full aperture.
264 allowdac Allow double-address cycle (DAC) mode, i.e. DMA >4GB.
265 DAC is used with 32-bit PCI to push a 64-bit address in
266 two cycles. When off all DMA over >4GB is forced through
267 an IOMMU or software bounce buffering.
268 nodac Forbid DAC mode, i.e. DMA >4GB.
269 panic Always panic when IOMMU overflows.
270 calgary Use the Calgary IOMMU if it is available
272 iommu options only relevant to the software bounce buffering (SWIOTLB) IOMMU
274 swiotlb=<pages>[,force]
275 <pages> Prereserve that many 128K pages for the software IO
277 force Force all IO through the software TLB.
279 Settings for the IBM Calgary hardware IOMMU currently found in IBM
280 pSeries and xSeries machines:
282 calgary=[64k,128k,256k,512k,1M,2M,4M,8M]
283 calgary=[translate_empty_slots]
284 calgary=[disable=<PCI bus number>]
285 panic Always panic when IOMMU overflows
287 64k,...,8M - Set the size of each PCI slot's translation table
288 when using the Calgary IOMMU. This is the size of the translation
289 table itself in main memory. The smallest table, 64k, covers an IO
290 space of 32MB; the largest, 8MB table, can cover an IO space of
291 4GB. Normally the kernel will make the right choice by itself.
293 translate_empty_slots - Enable translation even on slots that have
294 no devices attached to them, in case a device will be hotplugged
297 disable=<PCI bus number> - Disable translation on a given PHB. For
298 example, the built-in graphics adapter resides on the first bridge
299 (PCI bus number 0); if translation (isolation) is enabled on this
300 bridge, X servers that access the hardware directly from user
301 space might stop working. Use this option if you have devices that
302 are accessed from userspace directly on some PCI host bridge.
306 oops=panic Always panic on oopses. Default is to just kill the process,
307 but there is a small probability of deadlocking the machine.
308 This will also cause panics on machine check exceptions.
309 Useful together with panic=30 to trigger a reboot.
311 kstack=N Print N words from the kernel stack in oops dumps.
313 pagefaulttrace Dump all page faults. Only useful for extreme debugging
314 and will create a lot of output.
316 call_trace=[old|both|newfallback|new]
317 old: use old inexact backtracer
318 new: use new exact dwarf2 unwinder
319 both: print entries from both
320 newfallback: use new unwinder but fall back to old if it gets
326 Do not use GB pages for kernel direct mappings.
328 Use GB pages for kernel direct mappings.