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 on uncorrected errors, log corrected errors
18 1: panic or SIGBUS on uncorrected errors, log corrected errors
19 2: SIGBUS or log uncorrected errors, log corrected errors
20 3: never panic or SIGBUS, log all errors (for testing only)
22 Can be also set using sysfs which is preferable.
24 nomce (for compatibility with i386): same as mce=off
26 Everything else is in sysfs now.
30 apic Use IO-APIC. Default
32 noapic Don't use the IO-APIC.
34 disableapic Don't use the local APIC
36 nolapic Don't use the local APIC (alias for i386 compatibility)
38 pirq=... See Documentation/x86/i386/IO-APIC.txt
40 noapictimer Don't set up the APIC timer
42 no_timer_check Don't check the IO-APIC timer. This can work around
43 problems with incorrect timer initialization on some boards.
45 apicmaintimer Run time keeping from the local APIC timer instead
46 of using the PIT/HPET interrupt for this. This is useful
47 when the PIT/HPET interrupts are unreliable.
49 noapicmaintimer Don't do time keeping using the APIC timer.
50 Useful when this option was auto selected, but doesn't work.
53 Do APIC timer calibration using the pmtimer. Implies
54 apicmaintimer. Useful when your PIT timer is totally
59 syntax: earlyprintk=vga
60 earlyprintk=serial[,ttySn[,baudrate]]
62 The early console is useful when the kernel crashes before the
63 normal console is initialized. It is not enabled by
64 default because it has some cosmetic problems.
65 Append ,keep to not disable it when the real console takes over.
66 Only vga or serial at a time, not both.
67 Currently only ttyS0 and ttyS1 are supported.
68 Interaction with the standard serial driver is not very good.
69 The VGA output is eventually overwritten by the real console.
74 Don't use the CPU time stamp counter to read the wall time.
75 This can be used to work around timing problems on multiprocessor systems
76 with not properly synchronized CPUs.
79 Report when timer interrupts are lost because some code turned off
80 interrupts for too long.
83 Don't use the HPET timer.
88 Don't do power saving in the idle loop using HLT, but poll for rescheduling
89 event. This will make the CPUs eat a lot more power, but may be useful
90 to get slightly better performance in multiprocessor benchmarks. It also
91 makes some profiling using performance counters more accurate.
92 Please note that on systems with MONITOR/MWAIT support (like Intel EM64T
93 CPUs) this option has no performance advantage over the normal idle loop.
94 It may also interact badly with hyperthreading.
98 reboot=b[ios] | t[riple] | k[bd] | a[cpi] | e[fi] [, [w]arm | [c]old]
99 bios Use the CPU reboot vector for warm reset
100 warm Don't set the cold reboot flag
101 cold Set the cold reboot flag
102 triple Force a triple fault (init)
103 kbd Use the keyboard controller. cold reset (default)
104 acpi Use the ACPI RESET_REG in the FADT. If ACPI is not configured or the
105 ACPI reset does not work, the reboot path attempts the reset using
106 the keyboard controller.
107 efi Use efi reset_system runtime service. If EFI is not configured or the
108 EFI reset does not work, the reboot path attempts the reset using
109 the keyboard controller.
111 Using warm reset will be much faster especially on big memory
112 systems because the BIOS will not go through the memory check.
113 Disadvantage is that not all hardware will be completely reinitialized
114 on reboot so there may be boot problems on some systems.
118 Don't stop other CPUs on reboot. This can make reboot more reliable
121 Non Executable Mappings
130 additional_cpus=NUM Allow NUM more CPUs for hotplug
131 (defaults are specified by the BIOS, see Documentation/x86/x86_64/cpu-hotplug-spec)
135 numa=off Only set up a single NUMA node spanning all memory.
137 numa=noacpi Don't parse the SRAT table for NUMA setup
140 If a number, fakes CMDLINE nodes and ignores NUMA setup of the
141 actual machine. Otherwise, system memory is configured
142 depending on the sizes and coefficients listed. For example:
143 numa=fake=2*512,1024,4*256,*128
144 gives two 512M nodes, a 1024M node, four 256M nodes, and the
145 rest split into 128M chunks. If the last character of CMDLINE
146 is a *, the remaining memory is divided up equally among its
149 gives two 512M nodes and the rest split into two nodes.
150 Otherwise, the remaining system RAM is allocated to an
154 Only allow hotadd memory to preallocate page structures upto
155 percent of already available memory.
156 numa=hotadd=0 will disable hotadd memory.
160 acpi=off Don't enable ACPI
161 acpi=ht Use ACPI boot table parsing, but don't enable ACPI
163 acpi=force Force ACPI on (currently not needed)
165 acpi=strict Disable out of spec ACPI workarounds.
167 acpi_sci={edge,level,high,low} Set up ACPI SCI interrupt.
169 acpi=noirq Don't route interrupts
173 pci=off Don't use PCI
174 pci=conf1 Use conf1 access.
175 pci=conf2 Use conf2 access.
177 pci=assign-busses Assign busses
178 pci=irqmask=MASK Set PCI interrupt mask to MASK
179 pci=lastbus=NUMBER Scan upto NUMBER busses, no matter what the mptable says.
180 pci=noacpi Don't use ACPI to set up PCI interrupt routing.
182 IOMMU (input/output memory management unit)
184 Currently four x86-64 PCI-DMA mapping implementations exist:
186 1. <arch/x86_64/kernel/pci-nommu.c>: use no hardware/software IOMMU at all
187 (e.g. because you have < 3 GB memory).
188 Kernel boot message: "PCI-DMA: Disabling IOMMU"
190 2. <arch/x86_64/kernel/pci-gart.c>: AMD GART based hardware IOMMU.
191 Kernel boot message: "PCI-DMA: using GART IOMMU"
193 3. <arch/x86_64/kernel/pci-swiotlb.c> : Software IOMMU implementation. Used
194 e.g. if there is no hardware IOMMU in the system and it is need because
195 you have >3GB memory or told the kernel to us it (iommu=soft))
196 Kernel boot message: "PCI-DMA: Using software bounce buffering
199 4. <arch/x86_64/pci-calgary.c> : IBM Calgary hardware IOMMU. Used in IBM
200 pSeries and xSeries servers. This hardware IOMMU supports DMA address
201 mapping with memory protection, etc.
202 Kernel boot message: "PCI-DMA: Using Calgary IOMMU"
204 iommu=[<size>][,noagp][,off][,force][,noforce][,leak[=<nr_of_leak_pages>]
205 [,memaper[=<order>]][,merge][,forcesac][,fullflush][,nomerge]
206 [,noaperture][,calgary]
208 General iommu options:
209 off Don't initialize and use any kind of IOMMU.
210 noforce Don't force hardware IOMMU usage when it is not needed.
212 force Force the use of the hardware IOMMU even when it is
213 not actually needed (e.g. because < 3 GB memory).
214 soft Use software bounce buffering (SWIOTLB) (default for
215 Intel machines). This can be used to prevent the usage
216 of an available hardware IOMMU.
218 iommu options only relevant to the AMD GART hardware IOMMU:
219 <size> Set the size of the remapping area in bytes.
220 allowed Overwrite iommu off workarounds for specific chipsets.
221 fullflush Flush IOMMU on each allocation (default).
222 nofullflush Don't use IOMMU fullflush.
223 leak Turn on simple iommu leak tracing (only when
224 CONFIG_IOMMU_LEAK is on). Default number of leak pages
226 memaper[=<order>] Allocate an own aperture over RAM with size 32MB<<order.
227 (default: order=1, i.e. 64MB)
228 merge Do scatter-gather (SG) merging. Implies "force"
230 nomerge Don't do scatter-gather (SG) merging.
231 noaperture Ask the IOMMU not to touch the aperture for AGP.
232 forcesac Force single-address cycle (SAC) mode for masks <40bits
234 noagp Don't initialize the AGP driver and use full aperture.
235 allowdac Allow double-address cycle (DAC) mode, i.e. DMA >4GB.
236 DAC is used with 32-bit PCI to push a 64-bit address in
237 two cycles. When off all DMA over >4GB is forced through
238 an IOMMU or software bounce buffering.
239 nodac Forbid DAC mode, i.e. DMA >4GB.
240 panic Always panic when IOMMU overflows.
241 calgary Use the Calgary IOMMU if it is available
243 iommu options only relevant to the software bounce buffering (SWIOTLB) IOMMU
245 swiotlb=<pages>[,force]
246 <pages> Prereserve that many 128K pages for the software IO
248 force Force all IO through the software TLB.
250 Settings for the IBM Calgary hardware IOMMU currently found in IBM
251 pSeries and xSeries machines:
253 calgary=[64k,128k,256k,512k,1M,2M,4M,8M]
254 calgary=[translate_empty_slots]
255 calgary=[disable=<PCI bus number>]
256 panic Always panic when IOMMU overflows
258 64k,...,8M - Set the size of each PCI slot's translation table
259 when using the Calgary IOMMU. This is the size of the translation
260 table itself in main memory. The smallest table, 64k, covers an IO
261 space of 32MB; the largest, 8MB table, can cover an IO space of
262 4GB. Normally the kernel will make the right choice by itself.
264 translate_empty_slots - Enable translation even on slots that have
265 no devices attached to them, in case a device will be hotplugged
268 disable=<PCI bus number> - Disable translation on a given PHB. For
269 example, the built-in graphics adapter resides on the first bridge
270 (PCI bus number 0); if translation (isolation) is enabled on this
271 bridge, X servers that access the hardware directly from user
272 space might stop working. Use this option if you have devices that
273 are accessed from userspace directly on some PCI host bridge.
277 oops=panic Always panic on oopses. Default is to just kill the process,
278 but there is a small probability of deadlocking the machine.
279 This will also cause panics on machine check exceptions.
280 Useful together with panic=30 to trigger a reboot.
282 kstack=N Print N words from the kernel stack in oops dumps.
284 pagefaulttrace Dump all page faults. Only useful for extreme debugging
285 and will create a lot of output.
287 call_trace=[old|both|newfallback|new]
288 old: use old inexact backtracer
289 new: use new exact dwarf2 unwinder
290 both: print entries from both
291 newfallback: use new unwinder but fall back to old if it gets
297 Do not use GB pages for kernel direct mappings.
299 Use GB pages for kernel direct mappings.