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 mce=off disable machine check
9 mce=bootlog Enable logging of machine checks left over from booting.
10 Disabled by default on AMD because some BIOS leave bogus ones.
11 If your BIOS doesn't do that it's a good idea to enable though
12 to make sure you log even machine check events that result
13 in a reboot. On Intel systems it is enabled by default.
15 Disable boot machine check logging.
16 mce=tolerancelevel (number)
17 0: always panic, 1: panic if deadlock possible,
18 2: try to avoid panic, 3: never panic or exit (for testing)
20 Can be also set using sysfs which is preferable.
22 nomce (for compatibility with i386): same as mce=off
24 Everything else is in sysfs now.
28 apic Use IO-APIC. Default
30 noapic Don't use the IO-APIC.
32 disableapic Don't use the local APIC
34 nolapic Don't use the local APIC (alias for i386 compatibility)
36 pirq=... See Documentation/i386/IO-APIC.txt
38 noapictimer Don't set up the APIC timer
40 no_timer_check Don't check the IO-APIC timer. This can work around
41 problems with incorrect timer initialization on some boards.
43 apicmaintimer Run time keeping from the local APIC timer instead
44 of using the PIT/HPET interrupt for this. This is useful
45 when the PIT/HPET interrupts are unreliable.
47 noapicmaintimer Don't do time keeping using the APIC timer.
48 Useful when this option was auto selected, but doesn't work.
51 Do APIC timer calibration using the pmtimer. Implies
52 apicmaintimer. Useful when your PIT timer is totally
55 disable_8254_timer / enable_8254_timer
56 Enable interrupt 0 timer routing over the 8254 in addition to over
57 the IO-APIC. The kernel tries to set a sensible default.
61 syntax: earlyprintk=vga
62 earlyprintk=serial[,ttySn[,baudrate]]
64 The early console is useful when the kernel crashes before the
65 normal console is initialized. It is not enabled by
66 default because it has some cosmetic problems.
67 Append ,keep to not disable it when the real console takes over.
68 Only vga or serial at a time, not both.
69 Currently only ttyS0 and ttyS1 are supported.
70 Interaction with the standard serial driver is not very good.
71 The VGA output is eventually overwritten by the real console.
76 Don't use the CPU time stamp counter to read the wall time.
77 This can be used to work around timing problems on multiprocessor systems
78 with not properly synchronized CPUs.
81 Report when timer interrupts are lost because some code turned off
82 interrupts for too long.
84 nmi_watchdog=NUMBER[,panic]
86 0 don't use an NMI watchdog
87 1 use the IO-APIC timer for the NMI watchdog
88 2 use the local APIC for the NMI watchdog using a performance counter. Note
89 This will use one performance counter and the local APIC's performance
91 When panic is specified panic when an NMI watchdog timeout occurs.
92 This is useful when you use a panic=... timeout and need the box
96 Don't use the HPET timer.
101 Don't do power saving in the idle loop using HLT, but poll for rescheduling
102 event. This will make the CPUs eat a lot more power, but may be useful
103 to get slightly better performance in multiprocessor benchmarks. It also
104 makes some profiling using performance counters more accurate.
105 Please note that on systems with MONITOR/MWAIT support (like Intel EM64T
106 CPUs) this option has no performance advantage over the normal idle loop.
107 It may also interact badly with hyperthreading.
111 reboot=b[ios] | t[riple] | k[bd] [, [w]arm | [c]old]
112 bios Use the CPU reboot vector for warm reset
113 warm Don't set the cold reboot flag
114 cold Set the cold reboot flag
115 triple Force a triple fault (init)
116 kbd Use the keyboard controller. cold reset (default)
118 Using warm reset will be much faster especially on big memory
119 systems because the BIOS will not go through the memory check.
120 Disadvantage is that not all hardware will be completely reinitialized
121 on reboot so there may be boot problems on some systems.
125 Don't stop other CPUs on reboot. This can make reboot more reliable
128 Non Executable Mappings
137 nosmp Only use a single CPU
139 maxcpus=NUMBER only use upto NUMBER CPUs
141 cpumask=MASK only use cpus with bits set in mask
143 additional_cpus=NUM Allow NUM more CPUs for hotplug
144 (defaults are specified by the BIOS, see Documentation/x86_64/cpu-hotplug-spec)
148 numa=off Only set up a single NUMA node spanning all memory.
150 numa=noacpi Don't parse the SRAT table for NUMA setup
153 If a number, fakes CMDLINE nodes and ignores NUMA setup of the
154 actual machine. Otherwise, system memory is configured
155 depending on the sizes and coefficients listed. For example:
156 numa=fake=2*512,1024,4*256
157 gives two 512M nodes, a 1024M node, and four 256M nodes. The
158 remaining system RAM is allocated to an additional node.
161 Only allow hotadd memory to preallocate page structures upto
162 percent of already available memory.
163 numa=hotadd=0 will disable hotadd memory.
167 acpi=off Don't enable ACPI
168 acpi=ht Use ACPI boot table parsing, but don't enable ACPI
170 acpi=force Force ACPI on (currently not needed)
172 acpi=strict Disable out of spec ACPI workarounds.
174 acpi_sci={edge,level,high,low} Set up ACPI SCI interrupt.
176 acpi=noirq Don't route interrupts
180 pci=off Don't use PCI
181 pci=conf1 Use conf1 access.
182 pci=conf2 Use conf2 access.
184 pci=assign-busses Assign busses
185 pci=irqmask=MASK Set PCI interrupt mask to MASK
186 pci=lastbus=NUMBER Scan upto NUMBER busses, no matter what the mptable says.
187 pci=noacpi Don't use ACPI to set up PCI interrupt routing.
189 IOMMU (input/output memory management unit)
191 Currently four x86-64 PCI-DMA mapping implementations exist:
193 1. <arch/x86_64/kernel/pci-nommu.c>: use no hardware/software IOMMU at all
194 (e.g. because you have < 3 GB memory).
195 Kernel boot message: "PCI-DMA: Disabling IOMMU"
197 2. <arch/x86_64/kernel/pci-gart.c>: AMD GART based hardware IOMMU.
198 Kernel boot message: "PCI-DMA: using GART IOMMU"
200 3. <arch/x86_64/kernel/pci-swiotlb.c> : Software IOMMU implementation. Used
201 e.g. if there is no hardware IOMMU in the system and it is need because
202 you have >3GB memory or told the kernel to us it (iommu=soft))
203 Kernel boot message: "PCI-DMA: Using software bounce buffering
206 4. <arch/x86_64/pci-calgary.c> : IBM Calgary hardware IOMMU. Used in IBM
207 pSeries and xSeries servers. This hardware IOMMU supports DMA address
208 mapping with memory protection, etc.
209 Kernel boot message: "PCI-DMA: Using Calgary IOMMU"
211 iommu=[<size>][,noagp][,off][,force][,noforce][,leak[=<nr_of_leak_pages>]
212 [,memaper[=<order>]][,merge][,forcesac][,fullflush][,nomerge]
213 [,noaperture][,calgary]
215 General iommu options:
216 off Don't initialize and use any kind of IOMMU.
217 noforce Don't force hardware IOMMU usage when it is not needed.
219 force Force the use of the hardware IOMMU even when it is
220 not actually needed (e.g. because < 3 GB memory).
221 soft Use software bounce buffering (SWIOTLB) (default for
222 Intel machines). This can be used to prevent the usage
223 of an available hardware IOMMU.
225 iommu options only relevant to the AMD GART hardware IOMMU:
226 <size> Set the size of the remapping area in bytes.
227 allowed Overwrite iommu off workarounds for specific chipsets.
228 fullflush Flush IOMMU on each allocation (default).
229 nofullflush Don't use IOMMU fullflush.
230 leak Turn on simple iommu leak tracing (only when
231 CONFIG_IOMMU_LEAK is on). Default number of leak pages
233 memaper[=<order>] Allocate an own aperture over RAM with size 32MB<<order.
234 (default: order=1, i.e. 64MB)
235 merge Do scatter-gather (SG) merging. Implies "force"
237 nomerge Don't do scatter-gather (SG) merging.
238 noaperture Ask the IOMMU not to touch the aperture for AGP.
239 forcesac Force single-address cycle (SAC) mode for masks <40bits
241 noagp Don't initialize the AGP driver and use full aperture.
242 allowdac Allow double-address cycle (DAC) mode, i.e. DMA >4GB.
243 DAC is used with 32-bit PCI to push a 64-bit address in
244 two cycles. When off all DMA over >4GB is forced through
245 an IOMMU or software bounce buffering.
246 nodac Forbid DAC mode, i.e. DMA >4GB.
247 panic Always panic when IOMMU overflows.
248 calgary Use the Calgary IOMMU if it is available
250 iommu options only relevant to the software bounce buffering (SWIOTLB) IOMMU
252 swiotlb=<pages>[,force]
253 <pages> Prereserve that many 128K pages for the software IO
255 force Force all IO through the software TLB.
257 Settings for the IBM Calgary hardware IOMMU currently found in IBM
258 pSeries and xSeries machines:
260 calgary=[64k,128k,256k,512k,1M,2M,4M,8M]
261 calgary=[translate_empty_slots]
262 calgary=[disable=<PCI bus number>]
263 panic Always panic when IOMMU overflows
265 64k,...,8M - Set the size of each PCI slot's translation table
266 when using the Calgary IOMMU. This is the size of the translation
267 table itself in main memory. The smallest table, 64k, covers an IO
268 space of 32MB; the largest, 8MB table, can cover an IO space of
269 4GB. Normally the kernel will make the right choice by itself.
271 translate_empty_slots - Enable translation even on slots that have
272 no devices attached to them, in case a device will be hotplugged
275 disable=<PCI bus number> - Disable translation on a given PHB. For
276 example, the built-in graphics adapter resides on the first bridge
277 (PCI bus number 0); if translation (isolation) is enabled on this
278 bridge, X servers that access the hardware directly from user
279 space might stop working. Use this option if you have devices that
280 are accessed from userspace directly on some PCI host bridge.
284 oops=panic Always panic on oopses. Default is to just kill the process,
285 but there is a small probability of deadlocking the machine.
286 This will also cause panics on machine check exceptions.
287 Useful together with panic=30 to trigger a reboot.
289 kstack=N Print N words from the kernel stack in oops dumps.
291 pagefaulttrace Dump all page faults. Only useful for extreme debugging
292 and will create a lot of output.
294 call_trace=[old|both|newfallback|new]
295 old: use old inexact backtracer
296 new: use new exact dwarf2 unwinder
297 both: print entries from both
298 newfallback: use new unwinder but fall back to old if it gets