boot.rst 50 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442
  1. .. SPDX-License-Identifier: GPL-2.0
  2. ===========================
  3. The Linux/x86 Boot Protocol
  4. ===========================
  5. On the x86 platform, the Linux kernel uses a rather complicated boot
  6. convention. This has evolved partially due to historical aspects, as
  7. well as the desire in the early days to have the kernel itself be a
  8. bootable image, the complicated PC memory model and due to changed
  9. expectations in the PC industry caused by the effective demise of
  10. real-mode DOS as a mainstream operating system.
  11. Currently, the following versions of the Linux/x86 boot protocol exist.
  12. ============= ============================================================
  13. Old kernels zImage/Image support only. Some very early kernels
  14. may not even support a command line.
  15. Protocol 2.00 (Kernel 1.3.73) Added bzImage and initrd support, as
  16. well as a formalized way to communicate between the
  17. boot loader and the kernel. setup.S made relocatable,
  18. although the traditional setup area still assumed
  19. writable.
  20. Protocol 2.01 (Kernel 1.3.76) Added a heap overrun warning.
  21. Protocol 2.02 (Kernel 2.4.0-test3-pre3) New command line protocol.
  22. Lower the conventional memory ceiling. No overwrite
  23. of the traditional setup area, thus making booting
  24. safe for systems which use the EBDA from SMM or 32-bit
  25. BIOS entry points. zImage deprecated but still
  26. supported.
  27. Protocol 2.03 (Kernel 2.4.18-pre1) Explicitly makes the highest possible
  28. initrd address available to the bootloader.
  29. Protocol 2.04 (Kernel 2.6.14) Extend the syssize field to four bytes.
  30. Protocol 2.05 (Kernel 2.6.20) Make protected mode kernel relocatable.
  31. Introduce relocatable_kernel and kernel_alignment fields.
  32. Protocol 2.06 (Kernel 2.6.22) Added a field that contains the size of
  33. the boot command line.
  34. Protocol 2.07 (Kernel 2.6.24) Added paravirtualised boot protocol.
  35. Introduced hardware_subarch and hardware_subarch_data
  36. and KEEP_SEGMENTS flag in load_flags.
  37. Protocol 2.08 (Kernel 2.6.26) Added crc32 checksum and ELF format
  38. payload. Introduced payload_offset and payload_length
  39. fields to aid in locating the payload.
  40. Protocol 2.09 (Kernel 2.6.26) Added a field of 64-bit physical
  41. pointer to single linked list of struct setup_data.
  42. Protocol 2.10 (Kernel 2.6.31) Added a protocol for relaxed alignment
  43. beyond the kernel_alignment added, new init_size and
  44. pref_address fields. Added extended boot loader IDs.
  45. Protocol 2.11 (Kernel 3.6) Added a field for offset of EFI handover
  46. protocol entry point.
  47. Protocol 2.12 (Kernel 3.8) Added the xloadflags field and extension fields
  48. to struct boot_params for loading bzImage and ramdisk
  49. above 4G in 64bit.
  50. Protocol 2.13 (Kernel 3.14) Support 32- and 64-bit flags being set in
  51. xloadflags to support booting a 64-bit kernel from 32-bit
  52. EFI
  53. Protocol 2.14 BURNT BY INCORRECT COMMIT
  54. ae7e1238e68f2a472a125673ab506d49158c1889
  55. (x86/boot: Add ACPI RSDP address to setup_header)
  56. DO NOT USE!!! ASSUME SAME AS 2.13.
  57. Protocol 2.15 (Kernel 5.5) Added the kernel_info and kernel_info.setup_type_max.
  58. ============= ============================================================
  59. .. note::
  60. The protocol version number should be changed only if the setup header
  61. is changed. There is no need to update the version number if boot_params
  62. or kernel_info are changed. Additionally, it is recommended to use
  63. xloadflags (in this case the protocol version number should not be
  64. updated either) or kernel_info to communicate supported Linux kernel
  65. features to the boot loader. Due to very limited space available in
  66. the original setup header every update to it should be considered
  67. with great care. Starting from the protocol 2.15 the primary way to
  68. communicate things to the boot loader is the kernel_info.
  69. Memory Layout
  70. =============
  71. The traditional memory map for the kernel loader, used for Image or
  72. zImage kernels, typically looks like::
  73. | |
  74. 0A0000 +------------------------+
  75. | Reserved for BIOS | Do not use. Reserved for BIOS EBDA.
  76. 09A000 +------------------------+
  77. | Command line |
  78. | Stack/heap | For use by the kernel real-mode code.
  79. 098000 +------------------------+
  80. | Kernel setup | The kernel real-mode code.
  81. 090200 +------------------------+
  82. | Kernel boot sector | The kernel legacy boot sector.
  83. 090000 +------------------------+
  84. | Protected-mode kernel | The bulk of the kernel image.
  85. 010000 +------------------------+
  86. | Boot loader | <- Boot sector entry point 0000:7C00
  87. 001000 +------------------------+
  88. | Reserved for MBR/BIOS |
  89. 000800 +------------------------+
  90. | Typically used by MBR |
  91. 000600 +------------------------+
  92. | BIOS use only |
  93. 000000 +------------------------+
  94. When using bzImage, the protected-mode kernel was relocated to
  95. 0x100000 ("high memory"), and the kernel real-mode block (boot sector,
  96. setup, and stack/heap) was made relocatable to any address between
  97. 0x10000 and end of low memory. Unfortunately, in protocols 2.00 and
  98. 2.01 the 0x90000+ memory range is still used internally by the kernel;
  99. the 2.02 protocol resolves that problem.
  100. It is desirable to keep the "memory ceiling" -- the highest point in
  101. low memory touched by the boot loader -- as low as possible, since
  102. some newer BIOSes have begun to allocate some rather large amounts of
  103. memory, called the Extended BIOS Data Area, near the top of low
  104. memory. The boot loader should use the "INT 12h" BIOS call to verify
  105. how much low memory is available.
  106. Unfortunately, if INT 12h reports that the amount of memory is too
  107. low, there is usually nothing the boot loader can do but to report an
  108. error to the user. The boot loader should therefore be designed to
  109. take up as little space in low memory as it reasonably can. For
  110. zImage or old bzImage kernels, which need data written into the
  111. 0x90000 segment, the boot loader should make sure not to use memory
  112. above the 0x9A000 point; too many BIOSes will break above that point.
  113. For a modern bzImage kernel with boot protocol version >= 2.02, a
  114. memory layout like the following is suggested::
  115. ~ ~
  116. | Protected-mode kernel |
  117. 100000 +------------------------+
  118. | I/O memory hole |
  119. 0A0000 +------------------------+
  120. | Reserved for BIOS | Leave as much as possible unused
  121. ~ ~
  122. | Command line | (Can also be below the X+10000 mark)
  123. X+10000 +------------------------+
  124. | Stack/heap | For use by the kernel real-mode code.
  125. X+08000 +------------------------+
  126. | Kernel setup | The kernel real-mode code.
  127. | Kernel boot sector | The kernel legacy boot sector.
  128. X +------------------------+
  129. | Boot loader | <- Boot sector entry point 0000:7C00
  130. 001000 +------------------------+
  131. | Reserved for MBR/BIOS |
  132. 000800 +------------------------+
  133. | Typically used by MBR |
  134. 000600 +------------------------+
  135. | BIOS use only |
  136. 000000 +------------------------+
  137. ... where the address X is as low as the design of the boot loader permits.
  138. The Real-Mode Kernel Header
  139. ===========================
  140. In the following text, and anywhere in the kernel boot sequence, "a
  141. sector" refers to 512 bytes. It is independent of the actual sector
  142. size of the underlying medium.
  143. The first step in loading a Linux kernel should be to load the
  144. real-mode code (boot sector and setup code) and then examine the
  145. following header at offset 0x01f1. The real-mode code can total up to
  146. 32K, although the boot loader may choose to load only the first two
  147. sectors (1K) and then examine the bootup sector size.
  148. The header looks like:
  149. =========== ======== ===================== ============================================
  150. Offset/Size Proto Name Meaning
  151. =========== ======== ===================== ============================================
  152. 01F1/1 ALL(1) setup_sects The size of the setup in sectors
  153. 01F2/2 ALL root_flags If set, the root is mounted readonly
  154. 01F4/4 2.04+(2) syssize The size of the 32-bit code in 16-byte paras
  155. 01F8/2 ALL ram_size DO NOT USE - for bootsect.S use only
  156. 01FA/2 ALL vid_mode Video mode control
  157. 01FC/2 ALL root_dev Default root device number
  158. 01FE/2 ALL boot_flag 0xAA55 magic number
  159. 0200/2 2.00+ jump Jump instruction
  160. 0202/4 2.00+ header Magic signature "HdrS"
  161. 0206/2 2.00+ version Boot protocol version supported
  162. 0208/4 2.00+ realmode_swtch Boot loader hook (see below)
  163. 020C/2 2.00+ start_sys_seg The load-low segment (0x1000) (obsolete)
  164. 020E/2 2.00+ kernel_version Pointer to kernel version string
  165. 0210/1 2.00+ type_of_loader Boot loader identifier
  166. 0211/1 2.00+ loadflags Boot protocol option flags
  167. 0212/2 2.00+ setup_move_size Move to high memory size (used with hooks)
  168. 0214/4 2.00+ code32_start Boot loader hook (see below)
  169. 0218/4 2.00+ ramdisk_image initrd load address (set by boot loader)
  170. 021C/4 2.00+ ramdisk_size initrd size (set by boot loader)
  171. 0220/4 2.00+ bootsect_kludge DO NOT USE - for bootsect.S use only
  172. 0224/2 2.01+ heap_end_ptr Free memory after setup end
  173. 0226/1 2.02+(3) ext_loader_ver Extended boot loader version
  174. 0227/1 2.02+(3) ext_loader_type Extended boot loader ID
  175. 0228/4 2.02+ cmd_line_ptr 32-bit pointer to the kernel command line
  176. 022C/4 2.03+ initrd_addr_max Highest legal initrd address
  177. 0230/4 2.05+ kernel_alignment Physical addr alignment required for kernel
  178. 0234/1 2.05+ relocatable_kernel Whether kernel is relocatable or not
  179. 0235/1 2.10+ min_alignment Minimum alignment, as a power of two
  180. 0236/2 2.12+ xloadflags Boot protocol option flags
  181. 0238/4 2.06+ cmdline_size Maximum size of the kernel command line
  182. 023C/4 2.07+ hardware_subarch Hardware subarchitecture
  183. 0240/8 2.07+ hardware_subarch_data Subarchitecture-specific data
  184. 0248/4 2.08+ payload_offset Offset of kernel payload
  185. 024C/4 2.08+ payload_length Length of kernel payload
  186. 0250/8 2.09+ setup_data 64-bit physical pointer to linked list
  187. of struct setup_data
  188. 0258/8 2.10+ pref_address Preferred loading address
  189. 0260/4 2.10+ init_size Linear memory required during initialization
  190. 0264/4 2.11+ handover_offset Offset of handover entry point
  191. 0268/4 2.15+ kernel_info_offset Offset of the kernel_info
  192. =========== ======== ===================== ============================================
  193. .. note::
  194. (1) For backwards compatibility, if the setup_sects field contains 0, the
  195. real value is 4.
  196. (2) For boot protocol prior to 2.04, the upper two bytes of the syssize
  197. field are unusable, which means the size of a bzImage kernel
  198. cannot be determined.
  199. (3) Ignored, but safe to set, for boot protocols 2.02-2.09.
  200. If the "HdrS" (0x53726448) magic number is not found at offset 0x202,
  201. the boot protocol version is "old". Loading an old kernel, the
  202. following parameters should be assumed::
  203. Image type = zImage
  204. initrd not supported
  205. Real-mode kernel must be located at 0x90000.
  206. Otherwise, the "version" field contains the protocol version,
  207. e.g. protocol version 2.01 will contain 0x0201 in this field. When
  208. setting fields in the header, you must make sure only to set fields
  209. supported by the protocol version in use.
  210. Details of Header Fields
  211. ========================
  212. For each field, some are information from the kernel to the bootloader
  213. ("read"), some are expected to be filled out by the bootloader
  214. ("write"), and some are expected to be read and modified by the
  215. bootloader ("modify").
  216. All general purpose boot loaders should write the fields marked
  217. (obligatory). Boot loaders who want to load the kernel at a
  218. nonstandard address should fill in the fields marked (reloc); other
  219. boot loaders can ignore those fields.
  220. The byte order of all fields is littleendian (this is x86, after all.)
  221. ============ ===========
  222. Field name: setup_sects
  223. Type: read
  224. Offset/size: 0x1f1/1
  225. Protocol: ALL
  226. ============ ===========
  227. The size of the setup code in 512-byte sectors. If this field is
  228. 0, the real value is 4. The real-mode code consists of the boot
  229. sector (always one 512-byte sector) plus the setup code.
  230. ============ =================
  231. Field name: root_flags
  232. Type: modify (optional)
  233. Offset/size: 0x1f2/2
  234. Protocol: ALL
  235. ============ =================
  236. If this field is nonzero, the root defaults to readonly. The use of
  237. this field is deprecated; use the "ro" or "rw" options on the
  238. command line instead.
  239. ============ ===============================================
  240. Field name: syssize
  241. Type: read
  242. Offset/size: 0x1f4/4 (protocol 2.04+) 0x1f4/2 (protocol ALL)
  243. Protocol: 2.04+
  244. ============ ===============================================
  245. The size of the protected-mode code in units of 16-byte paragraphs.
  246. For protocol versions older than 2.04 this field is only two bytes
  247. wide, and therefore cannot be trusted for the size of a kernel if
  248. the LOAD_HIGH flag is set.
  249. ============ ===============
  250. Field name: ram_size
  251. Type: kernel internal
  252. Offset/size: 0x1f8/2
  253. Protocol: ALL
  254. ============ ===============
  255. This field is obsolete.
  256. ============ ===================
  257. Field name: vid_mode
  258. Type: modify (obligatory)
  259. Offset/size: 0x1fa/2
  260. ============ ===================
  261. Please see the section on SPECIAL COMMAND LINE OPTIONS.
  262. ============ =================
  263. Field name: root_dev
  264. Type: modify (optional)
  265. Offset/size: 0x1fc/2
  266. Protocol: ALL
  267. ============ =================
  268. The default root device device number. The use of this field is
  269. deprecated, use the "root=" option on the command line instead.
  270. ============ =========
  271. Field name: boot_flag
  272. Type: read
  273. Offset/size: 0x1fe/2
  274. Protocol: ALL
  275. ============ =========
  276. Contains 0xAA55. This is the closest thing old Linux kernels have
  277. to a magic number.
  278. ============ =======
  279. Field name: jump
  280. Type: read
  281. Offset/size: 0x200/2
  282. Protocol: 2.00+
  283. ============ =======
  284. Contains an x86 jump instruction, 0xEB followed by a signed offset
  285. relative to byte 0x202. This can be used to determine the size of
  286. the header.
  287. ============ =======
  288. Field name: header
  289. Type: read
  290. Offset/size: 0x202/4
  291. Protocol: 2.00+
  292. ============ =======
  293. Contains the magic number "HdrS" (0x53726448).
  294. ============ =======
  295. Field name: version
  296. Type: read
  297. Offset/size: 0x206/2
  298. Protocol: 2.00+
  299. ============ =======
  300. Contains the boot protocol version, in (major << 8)+minor format,
  301. e.g. 0x0204 for version 2.04, and 0x0a11 for a hypothetical version
  302. 10.17.
  303. ============ =================
  304. Field name: realmode_swtch
  305. Type: modify (optional)
  306. Offset/size: 0x208/4
  307. Protocol: 2.00+
  308. ============ =================
  309. Boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
  310. ============ =============
  311. Field name: start_sys_seg
  312. Type: read
  313. Offset/size: 0x20c/2
  314. Protocol: 2.00+
  315. ============ =============
  316. The load low segment (0x1000). Obsolete.
  317. ============ ==============
  318. Field name: kernel_version
  319. Type: read
  320. Offset/size: 0x20e/2
  321. Protocol: 2.00+
  322. ============ ==============
  323. If set to a nonzero value, contains a pointer to a NUL-terminated
  324. human-readable kernel version number string, less 0x200. This can
  325. be used to display the kernel version to the user. This value
  326. should be less than (0x200*setup_sects).
  327. For example, if this value is set to 0x1c00, the kernel version
  328. number string can be found at offset 0x1e00 in the kernel file.
  329. This is a valid value if and only if the "setup_sects" field
  330. contains the value 15 or higher, as::
  331. 0x1c00 < 15*0x200 (= 0x1e00) but
  332. 0x1c00 >= 14*0x200 (= 0x1c00)
  333. 0x1c00 >> 9 = 14, So the minimum value for setup_secs is 15.
  334. ============ ==================
  335. Field name: type_of_loader
  336. Type: write (obligatory)
  337. Offset/size: 0x210/1
  338. Protocol: 2.00+
  339. ============ ==================
  340. If your boot loader has an assigned id (see table below), enter
  341. 0xTV here, where T is an identifier for the boot loader and V is
  342. a version number. Otherwise, enter 0xFF here.
  343. For boot loader IDs above T = 0xD, write T = 0xE to this field and
  344. write the extended ID minus 0x10 to the ext_loader_type field.
  345. Similarly, the ext_loader_ver field can be used to provide more than
  346. four bits for the bootloader version.
  347. For example, for T = 0x15, V = 0x234, write::
  348. type_of_loader <- 0xE4
  349. ext_loader_type <- 0x05
  350. ext_loader_ver <- 0x23
  351. Assigned boot loader ids (hexadecimal):
  352. == =======================================
  353. 0 LILO
  354. (0x00 reserved for pre-2.00 bootloader)
  355. 1 Loadlin
  356. 2 bootsect-loader
  357. (0x20, all other values reserved)
  358. 3 Syslinux
  359. 4 Etherboot/gPXE/iPXE
  360. 5 ELILO
  361. 7 GRUB
  362. 8 U-Boot
  363. 9 Xen
  364. A Gujin
  365. B Qemu
  366. C Arcturus Networks uCbootloader
  367. D kexec-tools
  368. E Extended (see ext_loader_type)
  369. F Special (0xFF = undefined)
  370. 10 Reserved
  371. 11 Minimal Linux Bootloader
  372. <http://sebastian-plotz.blogspot.de>
  373. 12 OVMF UEFI virtualization stack
  374. == =======================================
  375. Please contact <hpa@zytor.com> if you need a bootloader ID value assigned.
  376. ============ ===================
  377. Field name: loadflags
  378. Type: modify (obligatory)
  379. Offset/size: 0x211/1
  380. Protocol: 2.00+
  381. ============ ===================
  382. This field is a bitmask.
  383. Bit 0 (read): LOADED_HIGH
  384. - If 0, the protected-mode code is loaded at 0x10000.
  385. - If 1, the protected-mode code is loaded at 0x100000.
  386. Bit 1 (kernel internal): KASLR_FLAG
  387. - Used internally by the compressed kernel to communicate
  388. KASLR status to kernel proper.
  389. - If 1, KASLR enabled.
  390. - If 0, KASLR disabled.
  391. Bit 5 (write): QUIET_FLAG
  392. - If 0, print early messages.
  393. - If 1, suppress early messages.
  394. This requests to the kernel (decompressor and early
  395. kernel) to not write early messages that require
  396. accessing the display hardware directly.
  397. Bit 6 (obsolete): KEEP_SEGMENTS
  398. Protocol: 2.07+
  399. - This flag is obsolete.
  400. Bit 7 (write): CAN_USE_HEAP
  401. Set this bit to 1 to indicate that the value entered in the
  402. heap_end_ptr is valid. If this field is clear, some setup code
  403. functionality will be disabled.
  404. ============ ===================
  405. Field name: setup_move_size
  406. Type: modify (obligatory)
  407. Offset/size: 0x212/2
  408. Protocol: 2.00-2.01
  409. ============ ===================
  410. When using protocol 2.00 or 2.01, if the real mode kernel is not
  411. loaded at 0x90000, it gets moved there later in the loading
  412. sequence. Fill in this field if you want additional data (such as
  413. the kernel command line) moved in addition to the real-mode kernel
  414. itself.
  415. The unit is bytes starting with the beginning of the boot sector.
  416. This field is can be ignored when the protocol is 2.02 or higher, or
  417. if the real-mode code is loaded at 0x90000.
  418. ============ ========================
  419. Field name: code32_start
  420. Type: modify (optional, reloc)
  421. Offset/size: 0x214/4
  422. Protocol: 2.00+
  423. ============ ========================
  424. The address to jump to in protected mode. This defaults to the load
  425. address of the kernel, and can be used by the boot loader to
  426. determine the proper load address.
  427. This field can be modified for two purposes:
  428. 1. as a boot loader hook (see Advanced Boot Loader Hooks below.)
  429. 2. if a bootloader which does not install a hook loads a
  430. relocatable kernel at a nonstandard address it will have to modify
  431. this field to point to the load address.
  432. ============ ==================
  433. Field name: ramdisk_image
  434. Type: write (obligatory)
  435. Offset/size: 0x218/4
  436. Protocol: 2.00+
  437. ============ ==================
  438. The 32-bit linear address of the initial ramdisk or ramfs. Leave at
  439. zero if there is no initial ramdisk/ramfs.
  440. ============ ==================
  441. Field name: ramdisk_size
  442. Type: write (obligatory)
  443. Offset/size: 0x21c/4
  444. Protocol: 2.00+
  445. ============ ==================
  446. Size of the initial ramdisk or ramfs. Leave at zero if there is no
  447. initial ramdisk/ramfs.
  448. ============ ===============
  449. Field name: bootsect_kludge
  450. Type: kernel internal
  451. Offset/size: 0x220/4
  452. Protocol: 2.00+
  453. ============ ===============
  454. This field is obsolete.
  455. ============ ==================
  456. Field name: heap_end_ptr
  457. Type: write (obligatory)
  458. Offset/size: 0x224/2
  459. Protocol: 2.01+
  460. ============ ==================
  461. Set this field to the offset (from the beginning of the real-mode
  462. code) of the end of the setup stack/heap, minus 0x0200.
  463. ============ ================
  464. Field name: ext_loader_ver
  465. Type: write (optional)
  466. Offset/size: 0x226/1
  467. Protocol: 2.02+
  468. ============ ================
  469. This field is used as an extension of the version number in the
  470. type_of_loader field. The total version number is considered to be
  471. (type_of_loader & 0x0f) + (ext_loader_ver << 4).
  472. The use of this field is boot loader specific. If not written, it
  473. is zero.
  474. Kernels prior to 2.6.31 did not recognize this field, but it is safe
  475. to write for protocol version 2.02 or higher.
  476. ============ =====================================================
  477. Field name: ext_loader_type
  478. Type: write (obligatory if (type_of_loader & 0xf0) == 0xe0)
  479. Offset/size: 0x227/1
  480. Protocol: 2.02+
  481. ============ =====================================================
  482. This field is used as an extension of the type number in
  483. type_of_loader field. If the type in type_of_loader is 0xE, then
  484. the actual type is (ext_loader_type + 0x10).
  485. This field is ignored if the type in type_of_loader is not 0xE.
  486. Kernels prior to 2.6.31 did not recognize this field, but it is safe
  487. to write for protocol version 2.02 or higher.
  488. ============ ==================
  489. Field name: cmd_line_ptr
  490. Type: write (obligatory)
  491. Offset/size: 0x228/4
  492. Protocol: 2.02+
  493. ============ ==================
  494. Set this field to the linear address of the kernel command line.
  495. The kernel command line can be located anywhere between the end of
  496. the setup heap and 0xA0000; it does not have to be located in the
  497. same 64K segment as the real-mode code itself.
  498. Fill in this field even if your boot loader does not support a
  499. command line, in which case you can point this to an empty string
  500. (or better yet, to the string "auto".) If this field is left at
  501. zero, the kernel will assume that your boot loader does not support
  502. the 2.02+ protocol.
  503. ============ ===============
  504. Field name: initrd_addr_max
  505. Type: read
  506. Offset/size: 0x22c/4
  507. Protocol: 2.03+
  508. ============ ===============
  509. The maximum address that may be occupied by the initial
  510. ramdisk/ramfs contents. For boot protocols 2.02 or earlier, this
  511. field is not present, and the maximum address is 0x37FFFFFF. (This
  512. address is defined as the address of the highest safe byte, so if
  513. your ramdisk is exactly 131072 bytes long and this field is
  514. 0x37FFFFFF, you can start your ramdisk at 0x37FE0000.)
  515. ============ ============================
  516. Field name: kernel_alignment
  517. Type: read/modify (reloc)
  518. Offset/size: 0x230/4
  519. Protocol: 2.05+ (read), 2.10+ (modify)
  520. ============ ============================
  521. Alignment unit required by the kernel (if relocatable_kernel is
  522. true.) A relocatable kernel that is loaded at an alignment
  523. incompatible with the value in this field will be realigned during
  524. kernel initialization.
  525. Starting with protocol version 2.10, this reflects the kernel
  526. alignment preferred for optimal performance; it is possible for the
  527. loader to modify this field to permit a lesser alignment. See the
  528. min_alignment and pref_address field below.
  529. ============ ==================
  530. Field name: relocatable_kernel
  531. Type: read (reloc)
  532. Offset/size: 0x234/1
  533. Protocol: 2.05+
  534. ============ ==================
  535. If this field is nonzero, the protected-mode part of the kernel can
  536. be loaded at any address that satisfies the kernel_alignment field.
  537. After loading, the boot loader must set the code32_start field to
  538. point to the loaded code, or to a boot loader hook.
  539. ============ =============
  540. Field name: min_alignment
  541. Type: read (reloc)
  542. Offset/size: 0x235/1
  543. Protocol: 2.10+
  544. ============ =============
  545. This field, if nonzero, indicates as a power of two the minimum
  546. alignment required, as opposed to preferred, by the kernel to boot.
  547. If a boot loader makes use of this field, it should update the
  548. kernel_alignment field with the alignment unit desired; typically::
  549. kernel_alignment = 1 << min_alignment
  550. There may be a considerable performance cost with an excessively
  551. misaligned kernel. Therefore, a loader should typically try each
  552. power-of-two alignment from kernel_alignment down to this alignment.
  553. ============ ==========
  554. Field name: xloadflags
  555. Type: read
  556. Offset/size: 0x236/2
  557. Protocol: 2.12+
  558. ============ ==========
  559. This field is a bitmask.
  560. Bit 0 (read): XLF_KERNEL_64
  561. - If 1, this kernel has the legacy 64-bit entry point at 0x200.
  562. Bit 1 (read): XLF_CAN_BE_LOADED_ABOVE_4G
  563. - If 1, kernel/boot_params/cmdline/ramdisk can be above 4G.
  564. Bit 2 (read): XLF_EFI_HANDOVER_32
  565. - If 1, the kernel supports the 32-bit EFI handoff entry point
  566. given at handover_offset.
  567. Bit 3 (read): XLF_EFI_HANDOVER_64
  568. - If 1, the kernel supports the 64-bit EFI handoff entry point
  569. given at handover_offset + 0x200.
  570. Bit 4 (read): XLF_EFI_KEXEC
  571. - If 1, the kernel supports kexec EFI boot with EFI runtime support.
  572. ============ ============
  573. Field name: cmdline_size
  574. Type: read
  575. Offset/size: 0x238/4
  576. Protocol: 2.06+
  577. ============ ============
  578. The maximum size of the command line without the terminating
  579. zero. This means that the command line can contain at most
  580. cmdline_size characters. With protocol version 2.05 and earlier, the
  581. maximum size was 255.
  582. ============ ====================================
  583. Field name: hardware_subarch
  584. Type: write (optional, defaults to x86/PC)
  585. Offset/size: 0x23c/4
  586. Protocol: 2.07+
  587. ============ ====================================
  588. In a paravirtualized environment the hardware low level architectural
  589. pieces such as interrupt handling, page table handling, and
  590. accessing process control registers needs to be done differently.
  591. This field allows the bootloader to inform the kernel we are in one
  592. one of those environments.
  593. ========== ==============================
  594. 0x00000000 The default x86/PC environment
  595. 0x00000001 lguest
  596. 0x00000002 Xen
  597. 0x00000003 Moorestown MID
  598. 0x00000004 CE4100 TV Platform
  599. ========== ==============================
  600. ============ =========================
  601. Field name: hardware_subarch_data
  602. Type: write (subarch-dependent)
  603. Offset/size: 0x240/8
  604. Protocol: 2.07+
  605. ============ =========================
  606. A pointer to data that is specific to hardware subarch
  607. This field is currently unused for the default x86/PC environment,
  608. do not modify.
  609. ============ ==============
  610. Field name: payload_offset
  611. Type: read
  612. Offset/size: 0x248/4
  613. Protocol: 2.08+
  614. ============ ==============
  615. If non-zero then this field contains the offset from the beginning
  616. of the protected-mode code to the payload.
  617. The payload may be compressed. The format of both the compressed and
  618. uncompressed data should be determined using the standard magic
  619. numbers. The currently supported compression formats are gzip
  620. (magic numbers 1F 8B or 1F 9E), bzip2 (magic number 42 5A), LZMA
  621. (magic number 5D 00), XZ (magic number FD 37), LZ4 (magic number
  622. 02 21) and ZSTD (magic number 28 B5). The uncompressed payload is
  623. currently always ELF (magic number 7F 45 4C 46).
  624. ============ ==============
  625. Field name: payload_length
  626. Type: read
  627. Offset/size: 0x24c/4
  628. Protocol: 2.08+
  629. ============ ==============
  630. The length of the payload.
  631. ============ ===============
  632. Field name: setup_data
  633. Type: write (special)
  634. Offset/size: 0x250/8
  635. Protocol: 2.09+
  636. ============ ===============
  637. The 64-bit physical pointer to NULL terminated single linked list of
  638. struct setup_data. This is used to define a more extensible boot
  639. parameters passing mechanism. The definition of struct setup_data is
  640. as follow::
  641. struct setup_data {
  642. u64 next;
  643. u32 type;
  644. u32 len;
  645. u8 data[0];
  646. };
  647. Where, the next is a 64-bit physical pointer to the next node of
  648. linked list, the next field of the last node is 0; the type is used
  649. to identify the contents of data; the len is the length of data
  650. field; the data holds the real payload.
  651. This list may be modified at a number of points during the bootup
  652. process. Therefore, when modifying this list one should always make
  653. sure to consider the case where the linked list already contains
  654. entries.
  655. The setup_data is a bit awkward to use for extremely large data objects,
  656. both because the setup_data header has to be adjacent to the data object
  657. and because it has a 32-bit length field. However, it is important that
  658. intermediate stages of the boot process have a way to identify which
  659. chunks of memory are occupied by kernel data.
  660. Thus setup_indirect struct and SETUP_INDIRECT type were introduced in
  661. protocol 2.15::
  662. struct setup_indirect {
  663. __u32 type;
  664. __u32 reserved; /* Reserved, must be set to zero. */
  665. __u64 len;
  666. __u64 addr;
  667. };
  668. The type member is a SETUP_INDIRECT | SETUP_* type. However, it cannot be
  669. SETUP_INDIRECT itself since making the setup_indirect a tree structure
  670. could require a lot of stack space in something that needs to parse it
  671. and stack space can be limited in boot contexts.
  672. Let's give an example how to point to SETUP_E820_EXT data using setup_indirect.
  673. In this case setup_data and setup_indirect will look like this::
  674. struct setup_data {
  675. __u64 next = 0 or <addr_of_next_setup_data_struct>;
  676. __u32 type = SETUP_INDIRECT;
  677. __u32 len = sizeof(setup_data);
  678. __u8 data[sizeof(setup_indirect)] = struct setup_indirect {
  679. __u32 type = SETUP_INDIRECT | SETUP_E820_EXT;
  680. __u32 reserved = 0;
  681. __u64 len = <len_of_SETUP_E820_EXT_data>;
  682. __u64 addr = <addr_of_SETUP_E820_EXT_data>;
  683. }
  684. }
  685. .. note::
  686. SETUP_INDIRECT | SETUP_NONE objects cannot be properly distinguished
  687. from SETUP_INDIRECT itself. So, this kind of objects cannot be provided
  688. by the bootloaders.
  689. ============ ============
  690. Field name: pref_address
  691. Type: read (reloc)
  692. Offset/size: 0x258/8
  693. Protocol: 2.10+
  694. ============ ============
  695. This field, if nonzero, represents a preferred load address for the
  696. kernel. A relocating bootloader should attempt to load at this
  697. address if possible.
  698. A non-relocatable kernel will unconditionally move itself and to run
  699. at this address.
  700. ============ =======
  701. Field name: init_size
  702. Type: read
  703. Offset/size: 0x260/4
  704. ============ =======
  705. This field indicates the amount of linear contiguous memory starting
  706. at the kernel runtime start address that the kernel needs before it
  707. is capable of examining its memory map. This is not the same thing
  708. as the total amount of memory the kernel needs to boot, but it can
  709. be used by a relocating boot loader to help select a safe load
  710. address for the kernel.
  711. The kernel runtime start address is determined by the following algorithm::
  712. if (relocatable_kernel)
  713. runtime_start = align_up(load_address, kernel_alignment)
  714. else
  715. runtime_start = pref_address
  716. ============ ===============
  717. Field name: handover_offset
  718. Type: read
  719. Offset/size: 0x264/4
  720. ============ ===============
  721. This field is the offset from the beginning of the kernel image to
  722. the EFI handover protocol entry point. Boot loaders using the EFI
  723. handover protocol to boot the kernel should jump to this offset.
  724. See EFI HANDOVER PROTOCOL below for more details.
  725. ============ ==================
  726. Field name: kernel_info_offset
  727. Type: read
  728. Offset/size: 0x268/4
  729. Protocol: 2.15+
  730. ============ ==================
  731. This field is the offset from the beginning of the kernel image to the
  732. kernel_info. The kernel_info structure is embedded in the Linux image
  733. in the uncompressed protected mode region.
  734. The kernel_info
  735. ===============
  736. The relationships between the headers are analogous to the various data
  737. sections:
  738. setup_header = .data
  739. boot_params/setup_data = .bss
  740. What is missing from the above list? That's right:
  741. kernel_info = .rodata
  742. We have been (ab)using .data for things that could go into .rodata or .bss for
  743. a long time, for lack of alternatives and -- especially early on -- inertia.
  744. Also, the BIOS stub is responsible for creating boot_params, so it isn't
  745. available to a BIOS-based loader (setup_data is, though).
  746. setup_header is permanently limited to 144 bytes due to the reach of the
  747. 2-byte jump field, which doubles as a length field for the structure, combined
  748. with the size of the "hole" in struct boot_params that a protected-mode loader
  749. or the BIOS stub has to copy it into. It is currently 119 bytes long, which
  750. leaves us with 25 very precious bytes. This isn't something that can be fixed
  751. without revising the boot protocol entirely, breaking backwards compatibility.
  752. boot_params proper is limited to 4096 bytes, but can be arbitrarily extended
  753. by adding setup_data entries. It cannot be used to communicate properties of
  754. the kernel image, because it is .bss and has no image-provided content.
  755. kernel_info solves this by providing an extensible place for information about
  756. the kernel image. It is readonly, because the kernel cannot rely on a
  757. bootloader copying its contents anywhere, but that is OK; if it becomes
  758. necessary it can still contain data items that an enabled bootloader would be
  759. expected to copy into a setup_data chunk.
  760. All kernel_info data should be part of this structure. Fixed size data have to
  761. be put before kernel_info_var_len_data label. Variable size data have to be put
  762. after kernel_info_var_len_data label. Each chunk of variable size data has to
  763. be prefixed with header/magic and its size, e.g.::
  764. kernel_info:
  765. .ascii "LToP" /* Header, Linux top (structure). */
  766. .long kernel_info_var_len_data - kernel_info
  767. .long kernel_info_end - kernel_info
  768. .long 0x01234567 /* Some fixed size data for the bootloaders. */
  769. kernel_info_var_len_data:
  770. example_struct: /* Some variable size data for the bootloaders. */
  771. .ascii "0123" /* Header/Magic. */
  772. .long example_struct_end - example_struct
  773. .ascii "Struct"
  774. .long 0x89012345
  775. example_struct_end:
  776. example_strings: /* Some variable size data for the bootloaders. */
  777. .ascii "ABCD" /* Header/Magic. */
  778. .long example_strings_end - example_strings
  779. .asciz "String_0"
  780. .asciz "String_1"
  781. example_strings_end:
  782. kernel_info_end:
  783. This way the kernel_info is self-contained blob.
  784. .. note::
  785. Each variable size data header/magic can be any 4-character string,
  786. without \0 at the end of the string, which does not collide with
  787. existing variable length data headers/magics.
  788. Details of the kernel_info Fields
  789. =================================
  790. ============ ========
  791. Field name: header
  792. Offset/size: 0x0000/4
  793. ============ ========
  794. Contains the magic number "LToP" (0x506f544c).
  795. ============ ========
  796. Field name: size
  797. Offset/size: 0x0004/4
  798. ============ ========
  799. This field contains the size of the kernel_info including kernel_info.header.
  800. It does not count kernel_info.kernel_info_var_len_data size. This field should be
  801. used by the bootloaders to detect supported fixed size fields in the kernel_info
  802. and beginning of kernel_info.kernel_info_var_len_data.
  803. ============ ========
  804. Field name: size_total
  805. Offset/size: 0x0008/4
  806. ============ ========
  807. This field contains the size of the kernel_info including kernel_info.header
  808. and kernel_info.kernel_info_var_len_data.
  809. ============ ==============
  810. Field name: setup_type_max
  811. Offset/size: 0x000c/4
  812. ============ ==============
  813. This field contains maximal allowed type for setup_data and setup_indirect structs.
  814. The Image Checksum
  815. ==================
  816. From boot protocol version 2.08 onwards the CRC-32 is calculated over
  817. the entire file using the characteristic polynomial 0x04C11DB7 and an
  818. initial remainder of 0xffffffff. The checksum is appended to the
  819. file; therefore the CRC of the file up to the limit specified in the
  820. syssize field of the header is always 0.
  821. The Kernel Command Line
  822. =======================
  823. The kernel command line has become an important way for the boot
  824. loader to communicate with the kernel. Some of its options are also
  825. relevant to the boot loader itself, see "special command line options"
  826. below.
  827. The kernel command line is a null-terminated string. The maximum
  828. length can be retrieved from the field cmdline_size. Before protocol
  829. version 2.06, the maximum was 255 characters. A string that is too
  830. long will be automatically truncated by the kernel.
  831. If the boot protocol version is 2.02 or later, the address of the
  832. kernel command line is given by the header field cmd_line_ptr (see
  833. above.) This address can be anywhere between the end of the setup
  834. heap and 0xA0000.
  835. If the protocol version is *not* 2.02 or higher, the kernel
  836. command line is entered using the following protocol:
  837. - At offset 0x0020 (word), "cmd_line_magic", enter the magic
  838. number 0xA33F.
  839. - At offset 0x0022 (word), "cmd_line_offset", enter the offset
  840. of the kernel command line (relative to the start of the
  841. real-mode kernel).
  842. - The kernel command line *must* be within the memory region
  843. covered by setup_move_size, so you may need to adjust this
  844. field.
  845. Memory Layout of The Real-Mode Code
  846. ===================================
  847. The real-mode code requires a stack/heap to be set up, as well as
  848. memory allocated for the kernel command line. This needs to be done
  849. in the real-mode accessible memory in bottom megabyte.
  850. It should be noted that modern machines often have a sizable Extended
  851. BIOS Data Area (EBDA). As a result, it is advisable to use as little
  852. of the low megabyte as possible.
  853. Unfortunately, under the following circumstances the 0x90000 memory
  854. segment has to be used:
  855. - When loading a zImage kernel ((loadflags & 0x01) == 0).
  856. - When loading a 2.01 or earlier boot protocol kernel.
  857. .. note::
  858. For the 2.00 and 2.01 boot protocols, the real-mode code
  859. can be loaded at another address, but it is internally
  860. relocated to 0x90000. For the "old" protocol, the
  861. real-mode code must be loaded at 0x90000.
  862. When loading at 0x90000, avoid using memory above 0x9a000.
  863. For boot protocol 2.02 or higher, the command line does not have to be
  864. located in the same 64K segment as the real-mode setup code; it is
  865. thus permitted to give the stack/heap the full 64K segment and locate
  866. the command line above it.
  867. The kernel command line should not be located below the real-mode
  868. code, nor should it be located in high memory.
  869. Sample Boot Configuartion
  870. =========================
  871. As a sample configuration, assume the following layout of the real
  872. mode segment.
  873. When loading below 0x90000, use the entire segment:
  874. ============= ===================
  875. 0x0000-0x7fff Real mode kernel
  876. 0x8000-0xdfff Stack and heap
  877. 0xe000-0xffff Kernel command line
  878. ============= ===================
  879. When loading at 0x90000 OR the protocol version is 2.01 or earlier:
  880. ============= ===================
  881. 0x0000-0x7fff Real mode kernel
  882. 0x8000-0x97ff Stack and heap
  883. 0x9800-0x9fff Kernel command line
  884. ============= ===================
  885. Such a boot loader should enter the following fields in the header::
  886. unsigned long base_ptr; /* base address for real-mode segment */
  887. if ( setup_sects == 0 ) {
  888. setup_sects = 4;
  889. }
  890. if ( protocol >= 0x0200 ) {
  891. type_of_loader = <type code>;
  892. if ( loading_initrd ) {
  893. ramdisk_image = <initrd_address>;
  894. ramdisk_size = <initrd_size>;
  895. }
  896. if ( protocol >= 0x0202 && loadflags & 0x01 )
  897. heap_end = 0xe000;
  898. else
  899. heap_end = 0x9800;
  900. if ( protocol >= 0x0201 ) {
  901. heap_end_ptr = heap_end - 0x200;
  902. loadflags |= 0x80; /* CAN_USE_HEAP */
  903. }
  904. if ( protocol >= 0x0202 ) {
  905. cmd_line_ptr = base_ptr + heap_end;
  906. strcpy(cmd_line_ptr, cmdline);
  907. } else {
  908. cmd_line_magic = 0xA33F;
  909. cmd_line_offset = heap_end;
  910. setup_move_size = heap_end + strlen(cmdline)+1;
  911. strcpy(base_ptr+cmd_line_offset, cmdline);
  912. }
  913. } else {
  914. /* Very old kernel */
  915. heap_end = 0x9800;
  916. cmd_line_magic = 0xA33F;
  917. cmd_line_offset = heap_end;
  918. /* A very old kernel MUST have its real-mode code
  919. loaded at 0x90000 */
  920. if ( base_ptr != 0x90000 ) {
  921. /* Copy the real-mode kernel */
  922. memcpy(0x90000, base_ptr, (setup_sects+1)*512);
  923. base_ptr = 0x90000; /* Relocated */
  924. }
  925. strcpy(0x90000+cmd_line_offset, cmdline);
  926. /* It is recommended to clear memory up to the 32K mark */
  927. memset(0x90000 + (setup_sects+1)*512, 0,
  928. (64-(setup_sects+1))*512);
  929. }
  930. Loading The Rest of The Kernel
  931. ==============================
  932. The 32-bit (non-real-mode) kernel starts at offset (setup_sects+1)*512
  933. in the kernel file (again, if setup_sects == 0 the real value is 4.)
  934. It should be loaded at address 0x10000 for Image/zImage kernels and
  935. 0x100000 for bzImage kernels.
  936. The kernel is a bzImage kernel if the protocol >= 2.00 and the 0x01
  937. bit (LOAD_HIGH) in the loadflags field is set::
  938. is_bzImage = (protocol >= 0x0200) && (loadflags & 0x01);
  939. load_address = is_bzImage ? 0x100000 : 0x10000;
  940. Note that Image/zImage kernels can be up to 512K in size, and thus use
  941. the entire 0x10000-0x90000 range of memory. This means it is pretty
  942. much a requirement for these kernels to load the real-mode part at
  943. 0x90000. bzImage kernels allow much more flexibility.
  944. Special Command Line Options
  945. ============================
  946. If the command line provided by the boot loader is entered by the
  947. user, the user may expect the following command line options to work.
  948. They should normally not be deleted from the kernel command line even
  949. though not all of them are actually meaningful to the kernel. Boot
  950. loader authors who need additional command line options for the boot
  951. loader itself should get them registered in
  952. Documentation/admin-guide/kernel-parameters.rst to make sure they will not
  953. conflict with actual kernel options now or in the future.
  954. vga=<mode>
  955. <mode> here is either an integer (in C notation, either
  956. decimal, octal, or hexadecimal) or one of the strings
  957. "normal" (meaning 0xFFFF), "ext" (meaning 0xFFFE) or "ask"
  958. (meaning 0xFFFD). This value should be entered into the
  959. vid_mode field, as it is used by the kernel before the command
  960. line is parsed.
  961. mem=<size>
  962. <size> is an integer in C notation optionally followed by
  963. (case insensitive) K, M, G, T, P or E (meaning << 10, << 20,
  964. << 30, << 40, << 50 or << 60). This specifies the end of
  965. memory to the kernel. This affects the possible placement of
  966. an initrd, since an initrd should be placed near end of
  967. memory. Note that this is an option to *both* the kernel and
  968. the bootloader!
  969. initrd=<file>
  970. An initrd should be loaded. The meaning of <file> is
  971. obviously bootloader-dependent, and some boot loaders
  972. (e.g. LILO) do not have such a command.
  973. In addition, some boot loaders add the following options to the
  974. user-specified command line:
  975. BOOT_IMAGE=<file>
  976. The boot image which was loaded. Again, the meaning of <file>
  977. is obviously bootloader-dependent.
  978. auto
  979. The kernel was booted without explicit user intervention.
  980. If these options are added by the boot loader, it is highly
  981. recommended that they are located *first*, before the user-specified
  982. or configuration-specified command line. Otherwise, "init=/bin/sh"
  983. gets confused by the "auto" option.
  984. Running the Kernel
  985. ==================
  986. The kernel is started by jumping to the kernel entry point, which is
  987. located at *segment* offset 0x20 from the start of the real mode
  988. kernel. This means that if you loaded your real-mode kernel code at
  989. 0x90000, the kernel entry point is 9020:0000.
  990. At entry, ds = es = ss should point to the start of the real-mode
  991. kernel code (0x9000 if the code is loaded at 0x90000), sp should be
  992. set up properly, normally pointing to the top of the heap, and
  993. interrupts should be disabled. Furthermore, to guard against bugs in
  994. the kernel, it is recommended that the boot loader sets fs = gs = ds =
  995. es = ss.
  996. In our example from above, we would do::
  997. /* Note: in the case of the "old" kernel protocol, base_ptr must
  998. be == 0x90000 at this point; see the previous sample code */
  999. seg = base_ptr >> 4;
  1000. cli(); /* Enter with interrupts disabled! */
  1001. /* Set up the real-mode kernel stack */
  1002. _SS = seg;
  1003. _SP = heap_end;
  1004. _DS = _ES = _FS = _GS = seg;
  1005. jmp_far(seg+0x20, 0); /* Run the kernel */
  1006. If your boot sector accesses a floppy drive, it is recommended to
  1007. switch off the floppy motor before running the kernel, since the
  1008. kernel boot leaves interrupts off and thus the motor will not be
  1009. switched off, especially if the loaded kernel has the floppy driver as
  1010. a demand-loaded module!
  1011. Advanced Boot Loader Hooks
  1012. ==========================
  1013. If the boot loader runs in a particularly hostile environment (such as
  1014. LOADLIN, which runs under DOS) it may be impossible to follow the
  1015. standard memory location requirements. Such a boot loader may use the
  1016. following hooks that, if set, are invoked by the kernel at the
  1017. appropriate time. The use of these hooks should probably be
  1018. considered an absolutely last resort!
  1019. IMPORTANT: All the hooks are required to preserve %esp, %ebp, %esi and
  1020. %edi across invocation.
  1021. realmode_swtch:
  1022. A 16-bit real mode far subroutine invoked immediately before
  1023. entering protected mode. The default routine disables NMI, so
  1024. your routine should probably do so, too.
  1025. code32_start:
  1026. A 32-bit flat-mode routine *jumped* to immediately after the
  1027. transition to protected mode, but before the kernel is
  1028. uncompressed. No segments, except CS, are guaranteed to be
  1029. set up (current kernels do, but older ones do not); you should
  1030. set them up to BOOT_DS (0x18) yourself.
  1031. After completing your hook, you should jump to the address
  1032. that was in this field before your boot loader overwrote it
  1033. (relocated, if appropriate.)
  1034. 32-bit Boot Protocol
  1035. ====================
  1036. For machine with some new BIOS other than legacy BIOS, such as EFI,
  1037. LinuxBIOS, etc, and kexec, the 16-bit real mode setup code in kernel
  1038. based on legacy BIOS can not be used, so a 32-bit boot protocol needs
  1039. to be defined.
  1040. In 32-bit boot protocol, the first step in loading a Linux kernel
  1041. should be to setup the boot parameters (struct boot_params,
  1042. traditionally known as "zero page"). The memory for struct boot_params
  1043. should be allocated and initialized to all zero. Then the setup header
  1044. from offset 0x01f1 of kernel image on should be loaded into struct
  1045. boot_params and examined. The end of setup header can be calculated as
  1046. follow::
  1047. 0x0202 + byte value at offset 0x0201
  1048. In addition to read/modify/write the setup header of the struct
  1049. boot_params as that of 16-bit boot protocol, the boot loader should
  1050. also fill the additional fields of the struct boot_params as
  1051. described in chapter :doc:`zero-page`.
  1052. After setting up the struct boot_params, the boot loader can load the
  1053. 32/64-bit kernel in the same way as that of 16-bit boot protocol.
  1054. In 32-bit boot protocol, the kernel is started by jumping to the
  1055. 32-bit kernel entry point, which is the start address of loaded
  1056. 32/64-bit kernel.
  1057. At entry, the CPU must be in 32-bit protected mode with paging
  1058. disabled; a GDT must be loaded with the descriptors for selectors
  1059. __BOOT_CS(0x10) and __BOOT_DS(0x18); both descriptors must be 4G flat
  1060. segment; __BOOT_CS must have execute/read permission, and __BOOT_DS
  1061. must have read/write permission; CS must be __BOOT_CS and DS, ES, SS
  1062. must be __BOOT_DS; interrupt must be disabled; %esi must hold the base
  1063. address of the struct boot_params; %ebp, %edi and %ebx must be zero.
  1064. 64-bit Boot Protocol
  1065. ====================
  1066. For machine with 64bit cpus and 64bit kernel, we could use 64bit bootloader
  1067. and we need a 64-bit boot protocol.
  1068. In 64-bit boot protocol, the first step in loading a Linux kernel
  1069. should be to setup the boot parameters (struct boot_params,
  1070. traditionally known as "zero page"). The memory for struct boot_params
  1071. could be allocated anywhere (even above 4G) and initialized to all zero.
  1072. Then, the setup header at offset 0x01f1 of kernel image on should be
  1073. loaded into struct boot_params and examined. The end of setup header
  1074. can be calculated as follows::
  1075. 0x0202 + byte value at offset 0x0201
  1076. In addition to read/modify/write the setup header of the struct
  1077. boot_params as that of 16-bit boot protocol, the boot loader should
  1078. also fill the additional fields of the struct boot_params as described
  1079. in chapter :doc:`zero-page`.
  1080. After setting up the struct boot_params, the boot loader can load
  1081. 64-bit kernel in the same way as that of 16-bit boot protocol, but
  1082. kernel could be loaded above 4G.
  1083. In 64-bit boot protocol, the kernel is started by jumping to the
  1084. 64-bit kernel entry point, which is the start address of loaded
  1085. 64-bit kernel plus 0x200.
  1086. At entry, the CPU must be in 64-bit mode with paging enabled.
  1087. The range with setup_header.init_size from start address of loaded
  1088. kernel and zero page and command line buffer get ident mapping;
  1089. a GDT must be loaded with the descriptors for selectors
  1090. __BOOT_CS(0x10) and __BOOT_DS(0x18); both descriptors must be 4G flat
  1091. segment; __BOOT_CS must have execute/read permission, and __BOOT_DS
  1092. must have read/write permission; CS must be __BOOT_CS and DS, ES, SS
  1093. must be __BOOT_DS; interrupt must be disabled; %rsi must hold the base
  1094. address of the struct boot_params.
  1095. EFI Handover Protocol (deprecated)
  1096. ==================================
  1097. This protocol allows boot loaders to defer initialisation to the EFI
  1098. boot stub. The boot loader is required to load the kernel/initrd(s)
  1099. from the boot media and jump to the EFI handover protocol entry point
  1100. which is hdr->handover_offset bytes from the beginning of
  1101. startup_{32,64}.
  1102. The boot loader MUST respect the kernel's PE/COFF metadata when it comes
  1103. to section alignment, the memory footprint of the executable image beyond
  1104. the size of the file itself, and any other aspect of the PE/COFF header
  1105. that may affect correct operation of the image as a PE/COFF binary in the
  1106. execution context provided by the EFI firmware.
  1107. The function prototype for the handover entry point looks like this::
  1108. efi_main(void *handle, efi_system_table_t *table, struct boot_params *bp)
  1109. 'handle' is the EFI image handle passed to the boot loader by the EFI
  1110. firmware, 'table' is the EFI system table - these are the first two
  1111. arguments of the "handoff state" as described in section 2.3 of the
  1112. UEFI specification. 'bp' is the boot loader-allocated boot params.
  1113. The boot loader *must* fill out the following fields in bp::
  1114. - hdr.cmd_line_ptr
  1115. - hdr.ramdisk_image (if applicable)
  1116. - hdr.ramdisk_size (if applicable)
  1117. All other fields should be zero.
  1118. NOTE: The EFI Handover Protocol is deprecated in favour of the ordinary PE/COFF
  1119. entry point, combined with the LINUX_EFI_INITRD_MEDIA_GUID based initrd
  1120. loading protocol (refer to [0] for an example of the bootloader side of
  1121. this), which removes the need for any knowledge on the part of the EFI
  1122. bootloader regarding the internal representation of boot_params or any
  1123. requirements/limitations regarding the placement of the command line
  1124. and ramdisk in memory, or the placement of the kernel image itself.
  1125. [0] https://github.com/u-boot/u-boot/commit/ec80b4735a593961fe701cc3a5d717d4739b0fd0