boot-options.txt 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295
  1. AMD64 specific boot options
  2. There are many others (usually documented in driver documentation), but
  3. only the AMD64 specific ones are listed here.
  4. Machine check
  5. mce=off disable machine check
  6. mce=bootlog Enable logging of machine checks left over from booting.
  7. Disabled by default on AMD because some BIOS leave bogus ones.
  8. If your BIOS doesn't do that it's a good idea to enable though
  9. to make sure you log even machine check events that result
  10. in a reboot. On Intel systems it is enabled by default.
  11. mce=nobootlog
  12. Disable boot machine check logging.
  13. mce=tolerancelevel (number)
  14. 0: always panic, 1: panic if deadlock possible,
  15. 2: try to avoid panic, 3: never panic or exit (for testing)
  16. default is 1
  17. Can be also set using sysfs which is preferable.
  18. nomce (for compatibility with i386): same as mce=off
  19. Everything else is in sysfs now.
  20. APICs
  21. apic Use IO-APIC. Default
  22. noapic Don't use the IO-APIC.
  23. disableapic Don't use the local APIC
  24. nolapic Don't use the local APIC (alias for i386 compatibility)
  25. pirq=... See Documentation/i386/IO-APIC.txt
  26. noapictimer Don't set up the APIC timer
  27. no_timer_check Don't check the IO-APIC timer. This can work around
  28. problems with incorrect timer initialization on some boards.
  29. apicmaintimer Run time keeping from the local APIC timer instead
  30. of using the PIT/HPET interrupt for this. This is useful
  31. when the PIT/HPET interrupts are unreliable.
  32. noapicmaintimer Don't do time keeping using the APIC timer.
  33. Useful when this option was auto selected, but doesn't work.
  34. apicpmtimer
  35. Do APIC timer calibration using the pmtimer. Implies
  36. apicmaintimer. Useful when your PIT timer is totally
  37. broken.
  38. disable_8254_timer / enable_8254_timer
  39. Enable interrupt 0 timer routing over the 8254 in addition to over
  40. the IO-APIC. The kernel tries to set a sensible default.
  41. Early Console
  42. syntax: earlyprintk=vga
  43. earlyprintk=serial[,ttySn[,baudrate]]
  44. The early console is useful when the kernel crashes before the
  45. normal console is initialized. It is not enabled by
  46. default because it has some cosmetic problems.
  47. Append ,keep to not disable it when the real console takes over.
  48. Only vga or serial at a time, not both.
  49. Currently only ttyS0 and ttyS1 are supported.
  50. Interaction with the standard serial driver is not very good.
  51. The VGA output is eventually overwritten by the real console.
  52. Timing
  53. notsc
  54. Don't use the CPU time stamp counter to read the wall time.
  55. This can be used to work around timing problems on multiprocessor systems
  56. with not properly synchronized CPUs.
  57. report_lost_ticks
  58. Report when timer interrupts are lost because some code turned off
  59. interrupts for too long.
  60. nmi_watchdog=NUMBER[,panic]
  61. NUMBER can be:
  62. 0 don't use an NMI watchdog
  63. 1 use the IO-APIC timer for the NMI watchdog
  64. 2 use the local APIC for the NMI watchdog using a performance counter. Note
  65. This will use one performance counter and the local APIC's performance
  66. vector.
  67. When panic is specified panic when an NMI watchdog timeout occurs.
  68. This is useful when you use a panic=... timeout and need the box
  69. quickly up again.
  70. nohpet
  71. Don't use the HPET timer.
  72. Idle loop
  73. idle=poll
  74. Don't do power saving in the idle loop using HLT, but poll for rescheduling
  75. event. This will make the CPUs eat a lot more power, but may be useful
  76. to get slightly better performance in multiprocessor benchmarks. It also
  77. makes some profiling using performance counters more accurate.
  78. Please note that on systems with MONITOR/MWAIT support (like Intel EM64T
  79. CPUs) this option has no performance advantage over the normal idle loop.
  80. It may also interact badly with hyperthreading.
  81. Rebooting
  82. reboot=b[ios] | t[riple] | k[bd] [, [w]arm | [c]old]
  83. bios Use the CPU reboot vector for warm reset
  84. warm Don't set the cold reboot flag
  85. cold Set the cold reboot flag
  86. triple Force a triple fault (init)
  87. kbd Use the keyboard controller. cold reset (default)
  88. Using warm reset will be much faster especially on big memory
  89. systems because the BIOS will not go through the memory check.
  90. Disadvantage is that not all hardware will be completely reinitialized
  91. on reboot so there may be boot problems on some systems.
  92. reboot=force
  93. Don't stop other CPUs on reboot. This can make reboot more reliable
  94. in some cases.
  95. Non Executable Mappings
  96. noexec=on|off
  97. on Enable(default)
  98. off Disable
  99. SMP
  100. nosmp Only use a single CPU
  101. maxcpus=NUMBER only use upto NUMBER CPUs
  102. cpumask=MASK only use cpus with bits set in mask
  103. additional_cpus=NUM Allow NUM more CPUs for hotplug
  104. (defaults are specified by the BIOS, see Documentation/x86_64/cpu-hotplug-spec)
  105. NUMA
  106. numa=off Only set up a single NUMA node spanning all memory.
  107. numa=noacpi Don't parse the SRAT table for NUMA setup
  108. numa=fake=X Fake X nodes and ignore NUMA setup of the actual machine.
  109. numa=hotadd=percent
  110. Only allow hotadd memory to preallocate page structures upto
  111. percent of already available memory.
  112. numa=hotadd=0 will disable hotadd memory.
  113. ACPI
  114. acpi=off Don't enable ACPI
  115. acpi=ht Use ACPI boot table parsing, but don't enable ACPI
  116. interpreter
  117. acpi=force Force ACPI on (currently not needed)
  118. acpi=strict Disable out of spec ACPI workarounds.
  119. acpi_sci={edge,level,high,low} Set up ACPI SCI interrupt.
  120. acpi=noirq Don't route interrupts
  121. PCI
  122. pci=off Don't use PCI
  123. pci=conf1 Use conf1 access.
  124. pci=conf2 Use conf2 access.
  125. pci=rom Assign ROMs.
  126. pci=assign-busses Assign busses
  127. pci=irqmask=MASK Set PCI interrupt mask to MASK
  128. pci=lastbus=NUMBER Scan upto NUMBER busses, no matter what the mptable says.
  129. pci=noacpi Don't use ACPI to set up PCI interrupt routing.
  130. IOMMU (input/output memory management unit)
  131. Currently four x86-64 PCI-DMA mapping implementations exist:
  132. 1. <arch/x86_64/kernel/pci-nommu.c>: use no hardware/software IOMMU at all
  133. (e.g. because you have < 3 GB memory).
  134. Kernel boot message: "PCI-DMA: Disabling IOMMU"
  135. 2. <arch/x86_64/kernel/pci-gart.c>: AMD GART based hardware IOMMU.
  136. Kernel boot message: "PCI-DMA: using GART IOMMU"
  137. 3. <arch/x86_64/kernel/pci-swiotlb.c> : Software IOMMU implementation. Used
  138. e.g. if there is no hardware IOMMU in the system and it is need because
  139. you have >3GB memory or told the kernel to us it (iommu=soft))
  140. Kernel boot message: "PCI-DMA: Using software bounce buffering
  141. for IO (SWIOTLB)"
  142. 4. <arch/x86_64/pci-calgary.c> : IBM Calgary hardware IOMMU. Used in IBM
  143. pSeries and xSeries servers. This hardware IOMMU supports DMA address
  144. mapping with memory protection, etc.
  145. Kernel boot message: "PCI-DMA: Using Calgary IOMMU"
  146. iommu=[<size>][,noagp][,off][,force][,noforce][,leak[=<nr_of_leak_pages>]
  147. [,memaper[=<order>]][,merge][,forcesac][,fullflush][,nomerge]
  148. [,noaperture][,calgary]
  149. General iommu options:
  150. off Don't initialize and use any kind of IOMMU.
  151. noforce Don't force hardware IOMMU usage when it is not needed.
  152. (default).
  153. force Force the use of the hardware IOMMU even when it is
  154. not actually needed (e.g. because < 3 GB memory).
  155. soft Use software bounce buffering (SWIOTLB) (default for
  156. Intel machines). This can be used to prevent the usage
  157. of an available hardware IOMMU.
  158. iommu options only relevant to the AMD GART hardware IOMMU:
  159. <size> Set the size of the remapping area in bytes.
  160. allowed Overwrite iommu off workarounds for specific chipsets.
  161. fullflush Flush IOMMU on each allocation (default).
  162. nofullflush Don't use IOMMU fullflush.
  163. leak Turn on simple iommu leak tracing (only when
  164. CONFIG_IOMMU_LEAK is on). Default number of leak pages
  165. is 20.
  166. memaper[=<order>] Allocate an own aperture over RAM with size 32MB<<order.
  167. (default: order=1, i.e. 64MB)
  168. merge Do scatter-gather (SG) merging. Implies "force"
  169. (experimental).
  170. nomerge Don't do scatter-gather (SG) merging.
  171. noaperture Ask the IOMMU not to touch the aperture for AGP.
  172. forcesac Force single-address cycle (SAC) mode for masks <40bits
  173. (experimental).
  174. noagp Don't initialize the AGP driver and use full aperture.
  175. allowdac Allow double-address cycle (DAC) mode, i.e. DMA >4GB.
  176. DAC is used with 32-bit PCI to push a 64-bit address in
  177. two cycles. When off all DMA over >4GB is forced through
  178. an IOMMU or software bounce buffering.
  179. nodac Forbid DAC mode, i.e. DMA >4GB.
  180. panic Always panic when IOMMU overflows.
  181. calgary Use the Calgary IOMMU if it is available
  182. iommu options only relevant to the software bounce buffering (SWIOTLB) IOMMU
  183. implementation:
  184. swiotlb=<pages>[,force]
  185. <pages> Prereserve that many 128K pages for the software IO
  186. bounce buffering.
  187. force Force all IO through the software TLB.
  188. Settings for the IBM Calgary hardware IOMMU currently found in IBM
  189. pSeries and xSeries machines:
  190. calgary=[64k,128k,256k,512k,1M,2M,4M,8M]
  191. calgary=[translate_empty_slots]
  192. calgary=[disable=<PCI bus number>]
  193. panic Always panic when IOMMU overflows
  194. 64k,...,8M - Set the size of each PCI slot's translation table
  195. when using the Calgary IOMMU. This is the size of the translation
  196. table itself in main memory. The smallest table, 64k, covers an IO
  197. space of 32MB; the largest, 8MB table, can cover an IO space of
  198. 4GB. Normally the kernel will make the right choice by itself.
  199. translate_empty_slots - Enable translation even on slots that have
  200. no devices attached to them, in case a device will be hotplugged
  201. in the future.
  202. disable=<PCI bus number> - Disable translation on a given PHB. For
  203. example, the built-in graphics adapter resides on the first bridge
  204. (PCI bus number 0); if translation (isolation) is enabled on this
  205. bridge, X servers that access the hardware directly from user
  206. space might stop working. Use this option if you have devices that
  207. are accessed from userspace directly on some PCI host bridge.
  208. Debugging
  209. oops=panic Always panic on oopses. Default is to just kill the process,
  210. but there is a small probability of deadlocking the machine.
  211. This will also cause panics on machine check exceptions.
  212. Useful together with panic=30 to trigger a reboot.
  213. kstack=N Print N words from the kernel stack in oops dumps.
  214. pagefaulttrace Dump all page faults. Only useful for extreme debugging
  215. and will create a lot of output.
  216. call_trace=[old|both|newfallback|new]
  217. old: use old inexact backtracer
  218. new: use new exact dwarf2 unwinder
  219. both: print entries from both
  220. newfallback: use new unwinder but fall back to old if it gets
  221. stuck (default)
  222. Miscellaneous