1 THE LINUX/x86 BOOT PROTOCOL
2 ---------------------------
4 On the x86 platform, the Linux kernel uses a rather complicated boot
5 convention. This has evolved partially due to historical aspects, as
6 well as the desire in the early days to have the kernel itself be a
7 bootable image, the complicated PC memory model and due to changed
8 expectations in the PC industry caused by the effective demise of
9 real-mode DOS as a mainstream operating system.
11 Currently, the following versions of the Linux/x86 boot protocol exist.
13 Old kernels: zImage/Image support only. Some very early kernels
14 may not even support a command line.
16 Protocol 2.00: (Kernel 1.3.73) Added bzImage and initrd support, as
17 well as a formalized way to communicate between the
18 boot loader and the kernel. setup.S made relocatable,
19 although the traditional setup area still assumed
22 Protocol 2.01: (Kernel 1.3.76) Added a heap overrun warning.
24 Protocol 2.02: (Kernel 2.4.0-test3-pre3) New command line protocol.
25 Lower the conventional memory ceiling. No overwrite
26 of the traditional setup area, thus making booting
27 safe for systems which use the EBDA from SMM or 32-bit
28 BIOS entry points. zImage deprecated but still
31 Protocol 2.03: (Kernel 2.4.18-pre1) Explicitly makes the highest possible
32 initrd address available to the bootloader.
34 Protocol 2.04: (Kernel 2.6.14) Extend the syssize field to four bytes.
36 Protocol 2.05: (Kernel 2.6.20) Make protected mode kernel relocatable.
37 Introduce relocatable_kernel and kernel_alignment fields.
39 Protocol 2.06: (Kernel 2.6.22) Added a field that contains the size of
40 the boot command line.
42 Protocol 2.07: (Kernel 2.6.24) Added paravirtualised boot protocol.
43 Introduced hardware_subarch and hardware_subarch_data
44 and KEEP_SEGMENTS flag in load_flags.
46 Protocol 2.08: (Kernel 2.6.26) Added crc32 checksum and ELF format
47 payload. Introduced payload_offset and payload_length
48 fields to aid in locating the payload.
50 Protocol 2.09: (Kernel 2.6.26) Added a field of 64-bit physical
51 pointer to single linked list of struct setup_data.
53 Protocol 2.10: (Kernel 2.6.31) Added a protocol for relaxed alignment
54 beyond the kernel_alignment added, new init_size and
55 pref_address fields. Added extended boot loader IDs.
57 Protocol 2.11: (Kernel 3.6) Added a field for offset of EFI handover
60 Protocol 2.12: (Kernel 3.8) Added the xloadflags field and extension fields
61 to struct boot_params for loading bzImage and ramdisk
64 Protocol 2.13: (Kernel 3.14) Support 32- and 64-bit flags being set in
65 xloadflags to support booting a 64-bit kernel from 32-bit
68 Protocol 2.14: (Kernel 4.20) Added acpi_rsdp_addr holding the physical
69 address of the ACPI RSDP table.
70 The bootloader updates version with:
71 0x8000 | min(kernel-version, bootloader-version)
72 kernel-version being the protocol version supported by
73 the kernel and bootloader-version the protocol version
74 supported by the bootloader.
78 The traditional memory map for the kernel loader, used for Image or
79 zImage kernels, typically looks like:
82 0A0000 +------------------------+
83 | Reserved for BIOS | Do not use. Reserved for BIOS EBDA.
84 09A000 +------------------------+
86 | Stack/heap | For use by the kernel real-mode code.
87 098000 +------------------------+
88 | Kernel setup | The kernel real-mode code.
89 090200 +------------------------+
90 | Kernel boot sector | The kernel legacy boot sector.
91 090000 +------------------------+
92 | Protected-mode kernel | The bulk of the kernel image.
93 010000 +------------------------+
94 | Boot loader | <- Boot sector entry point 0000:7C00
95 001000 +------------------------+
96 | Reserved for MBR/BIOS |
97 000800 +------------------------+
98 | Typically used by MBR |
99 000600 +------------------------+
101 000000 +------------------------+
104 When using bzImage, the protected-mode kernel was relocated to
105 0x100000 ("high memory"), and the kernel real-mode block (boot sector,
106 setup, and stack/heap) was made relocatable to any address between
107 0x10000 and end of low memory. Unfortunately, in protocols 2.00 and
108 2.01 the 0x90000+ memory range is still used internally by the kernel;
109 the 2.02 protocol resolves that problem.
111 It is desirable to keep the "memory ceiling" -- the highest point in
112 low memory touched by the boot loader -- as low as possible, since
113 some newer BIOSes have begun to allocate some rather large amounts of
114 memory, called the Extended BIOS Data Area, near the top of low
115 memory. The boot loader should use the "INT 12h" BIOS call to verify
116 how much low memory is available.
118 Unfortunately, if INT 12h reports that the amount of memory is too
119 low, there is usually nothing the boot loader can do but to report an
120 error to the user. The boot loader should therefore be designed to
121 take up as little space in low memory as it reasonably can. For
122 zImage or old bzImage kernels, which need data written into the
123 0x90000 segment, the boot loader should make sure not to use memory
124 above the 0x9A000 point; too many BIOSes will break above that point.
126 For a modern bzImage kernel with boot protocol version >= 2.02, a
127 memory layout like the following is suggested:
130 | Protected-mode kernel |
131 100000 +------------------------+
133 0A0000 +------------------------+
134 | Reserved for BIOS | Leave as much as possible unused
136 | Command line | (Can also be below the X+10000 mark)
137 X+10000 +------------------------+
138 | Stack/heap | For use by the kernel real-mode code.
139 X+08000 +------------------------+
140 | Kernel setup | The kernel real-mode code.
141 | Kernel boot sector | The kernel legacy boot sector.
142 X +------------------------+
143 | Boot loader | <- Boot sector entry point 0000:7C00
144 001000 +------------------------+
145 | Reserved for MBR/BIOS |
146 000800 +------------------------+
147 | Typically used by MBR |
148 000600 +------------------------+
150 000000 +------------------------+
152 ... where the address X is as low as the design of the boot loader
156 **** THE REAL-MODE KERNEL HEADER
158 In the following text, and anywhere in the kernel boot sequence, "a
159 sector" refers to 512 bytes. It is independent of the actual sector
160 size of the underlying medium.
162 The first step in loading a Linux kernel should be to load the
163 real-mode code (boot sector and setup code) and then examine the
164 following header at offset 0x01f1. The real-mode code can total up to
165 32K, although the boot loader may choose to load only the first two
166 sectors (1K) and then examine the bootup sector size.
168 The header looks like:
170 Offset Proto Name Meaning
173 01F1/1 ALL(1 setup_sects The size of the setup in sectors
174 01F2/2 ALL root_flags If set, the root is mounted readonly
175 01F4/4 2.04+(2 syssize The size of the 32-bit code in 16-byte paras
176 01F8/2 ALL ram_size DO NOT USE - for bootsect.S use only
177 01FA/2 ALL vid_mode Video mode control
178 01FC/2 ALL root_dev Default root device number
179 01FE/2 ALL boot_flag 0xAA55 magic number
180 0200/2 2.00+ jump Jump instruction
181 0202/4 2.00+ header Magic signature "HdrS"
182 0206/2 2.00+ version Boot protocol version supported
183 0208/4 2.00+ realmode_swtch Boot loader hook (see below)
184 020C/2 2.00+ start_sys_seg The load-low segment (0x1000) (obsolete)
185 020E/2 2.00+ kernel_version Pointer to kernel version string
186 0210/1 2.00+ type_of_loader Boot loader identifier
187 0211/1 2.00+ loadflags Boot protocol option flags
188 0212/2 2.00+ setup_move_size Move to high memory size (used with hooks)
189 0214/4 2.00+ code32_start Boot loader hook (see below)
190 0218/4 2.00+ ramdisk_image initrd load address (set by boot loader)
191 021C/4 2.00+ ramdisk_size initrd size (set by boot loader)
192 0220/4 2.00+ bootsect_kludge DO NOT USE - for bootsect.S use only
193 0224/2 2.01+ heap_end_ptr Free memory after setup end
194 0226/1 2.02+(3 ext_loader_ver Extended boot loader version
195 0227/1 2.02+(3 ext_loader_type Extended boot loader ID
196 0228/4 2.02+ cmd_line_ptr 32-bit pointer to the kernel command line
197 022C/4 2.03+ initrd_addr_max Highest legal initrd address
198 0230/4 2.05+ kernel_alignment Physical addr alignment required for kernel
199 0234/1 2.05+ relocatable_kernel Whether kernel is relocatable or not
200 0235/1 2.10+ min_alignment Minimum alignment, as a power of two
201 0236/2 2.12+ xloadflags Boot protocol option flags
202 0238/4 2.06+ cmdline_size Maximum size of the kernel command line
203 023C/4 2.07+ hardware_subarch Hardware subarchitecture
204 0240/8 2.07+ hardware_subarch_data Subarchitecture-specific data
205 0248/4 2.08+ payload_offset Offset of kernel payload
206 024C/4 2.08+ payload_length Length of kernel payload
207 0250/8 2.09+ setup_data 64-bit physical pointer to linked list
209 0258/8 2.10+ pref_address Preferred loading address
210 0260/4 2.10+ init_size Linear memory required during initialization
211 0264/4 2.11+ handover_offset Offset of handover entry point
212 0268/8 2.14+ acpi_rsdp_addr Physical address of RSDP table
214 (1) For backwards compatibility, if the setup_sects field contains 0, the
217 (2) For boot protocol prior to 2.04, the upper two bytes of the syssize
218 field are unusable, which means the size of a bzImage kernel
219 cannot be determined.
221 (3) Ignored, but safe to set, for boot protocols 2.02-2.09.
223 If the "HdrS" (0x53726448) magic number is not found at offset 0x202,
224 the boot protocol version is "old". Loading an old kernel, the
225 following parameters should be assumed:
229 Real-mode kernel must be located at 0x90000.
231 Otherwise, the "version" field contains the protocol version,
232 e.g. protocol version 2.01 will contain 0x0201 in this field. When
233 setting fields in the header, you must make sure only to set fields
234 supported by the protocol version in use.
237 **** DETAILS OF HEADER FIELDS
239 For each field, some are information from the kernel to the bootloader
240 ("read"), some are expected to be filled out by the bootloader
241 ("write"), and some are expected to be read and modified by the
242 bootloader ("modify").
244 All general purpose boot loaders should write the fields marked
245 (obligatory). Boot loaders who want to load the kernel at a
246 nonstandard address should fill in the fields marked (reloc); other
247 boot loaders can ignore those fields.
249 The byte order of all fields is littleendian (this is x86, after all.)
251 Field name: setup_sects
256 The size of the setup code in 512-byte sectors. If this field is
257 0, the real value is 4. The real-mode code consists of the boot
258 sector (always one 512-byte sector) plus the setup code.
260 Field name: root_flags
261 Type: modify (optional)
265 If this field is nonzero, the root defaults to readonly. The use of
266 this field is deprecated; use the "ro" or "rw" options on the
267 command line instead.
271 Offset/size: 0x1f4/4 (protocol 2.04+) 0x1f4/2 (protocol ALL)
274 The size of the protected-mode code in units of 16-byte paragraphs.
275 For protocol versions older than 2.04 this field is only two bytes
276 wide, and therefore cannot be trusted for the size of a kernel if
277 the LOAD_HIGH flag is set.
280 Type: kernel internal
284 This field is obsolete.
287 Type: modify (obligatory)
290 Please see the section on SPECIAL COMMAND LINE OPTIONS.
293 Type: modify (optional)
297 The default root device device number. The use of this field is
298 deprecated, use the "root=" option on the command line instead.
300 Field name: boot_flag
305 Contains 0xAA55. This is the closest thing old Linux kernels have
313 Contains an x86 jump instruction, 0xEB followed by a signed offset
314 relative to byte 0x202. This can be used to determine the size of
322 Contains the magic number "HdrS" (0x53726448).
329 Contains the boot protocol version, in (major << 8)+minor format,
330 e.g. 0x0204 for version 2.04, and 0x0a11 for a hypothetical version
333 Up to protocol version 2.13 this information is only read by the
334 bootloader. From protocol version 2.14 onwards the bootloader will
335 write the used protocol version or-ed with 0x8000 to the field. The
336 used protocol version will be the minimum of the supported protocol
337 versions of the bootloader and the kernel.
339 Field name: realmode_swtch
340 Type: modify (optional)
344 Boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
346 Field name: start_sys_seg
351 The load low segment (0x1000). Obsolete.
353 Field name: kernel_version
358 If set to a nonzero value, contains a pointer to a NUL-terminated
359 human-readable kernel version number string, less 0x200. This can
360 be used to display the kernel version to the user. This value
361 should be less than (0x200*setup_sects).
363 For example, if this value is set to 0x1c00, the kernel version
364 number string can be found at offset 0x1e00 in the kernel file.
365 This is a valid value if and only if the "setup_sects" field
366 contains the value 15 or higher, as:
368 0x1c00 < 15*0x200 (= 0x1e00) but
369 0x1c00 >= 14*0x200 (= 0x1c00)
371 0x1c00 >> 9 = 14, so the minimum value for setup_secs is 15.
373 Field name: type_of_loader
374 Type: write (obligatory)
378 If your boot loader has an assigned id (see table below), enter
379 0xTV here, where T is an identifier for the boot loader and V is
380 a version number. Otherwise, enter 0xFF here.
382 For boot loader IDs above T = 0xD, write T = 0xE to this field and
383 write the extended ID minus 0x10 to the ext_loader_type field.
384 Similarly, the ext_loader_ver field can be used to provide more than
385 four bits for the bootloader version.
387 For example, for T = 0x15, V = 0x234, write:
389 type_of_loader <- 0xE4
390 ext_loader_type <- 0x05
391 ext_loader_ver <- 0x23
393 Assigned boot loader ids (hexadecimal):
395 0 LILO (0x00 reserved for pre-2.00 bootloader)
397 2 bootsect-loader (0x20, all other values reserved)
399 4 Etherboot/gPXE/iPXE
406 C Arcturus Networks uCbootloader
408 E Extended (see ext_loader_type)
409 F Special (0xFF = undefined)
411 11 Minimal Linux Bootloader <http://sebastian-plotz.blogspot.de>
412 12 OVMF UEFI virtualization stack
414 Please contact <hpa@zytor.com> if you need a bootloader ID
417 Field name: loadflags
418 Type: modify (obligatory)
422 This field is a bitmask.
424 Bit 0 (read): LOADED_HIGH
425 - If 0, the protected-mode code is loaded at 0x10000.
426 - If 1, the protected-mode code is loaded at 0x100000.
428 Bit 1 (kernel internal): KASLR_FLAG
429 - Used internally by the compressed kernel to communicate
430 KASLR status to kernel proper.
432 If 0, KASLR disabled.
434 Bit 5 (write): QUIET_FLAG
435 - If 0, print early messages.
436 - If 1, suppress early messages.
437 This requests to the kernel (decompressor and early
438 kernel) to not write early messages that require
439 accessing the display hardware directly.
441 Bit 6 (write): KEEP_SEGMENTS
443 - If 0, reload the segment registers in the 32bit entry point.
444 - If 1, do not reload the segment registers in the 32bit entry point.
445 Assume that %cs %ds %ss %es are all set to flat segments with
446 a base of 0 (or the equivalent for their environment).
448 Bit 7 (write): CAN_USE_HEAP
449 Set this bit to 1 to indicate that the value entered in the
450 heap_end_ptr is valid. If this field is clear, some setup code
451 functionality will be disabled.
453 Field name: setup_move_size
454 Type: modify (obligatory)
458 When using protocol 2.00 or 2.01, if the real mode kernel is not
459 loaded at 0x90000, it gets moved there later in the loading
460 sequence. Fill in this field if you want additional data (such as
461 the kernel command line) moved in addition to the real-mode kernel
464 The unit is bytes starting with the beginning of the boot sector.
466 This field is can be ignored when the protocol is 2.02 or higher, or
467 if the real-mode code is loaded at 0x90000.
469 Field name: code32_start
470 Type: modify (optional, reloc)
474 The address to jump to in protected mode. This defaults to the load
475 address of the kernel, and can be used by the boot loader to
476 determine the proper load address.
478 This field can be modified for two purposes:
480 1. as a boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
482 2. if a bootloader which does not install a hook loads a
483 relocatable kernel at a nonstandard address it will have to modify
484 this field to point to the load address.
486 Field name: ramdisk_image
487 Type: write (obligatory)
491 The 32-bit linear address of the initial ramdisk or ramfs. Leave at
492 zero if there is no initial ramdisk/ramfs.
494 Field name: ramdisk_size
495 Type: write (obligatory)
499 Size of the initial ramdisk or ramfs. Leave at zero if there is no
500 initial ramdisk/ramfs.
502 Field name: bootsect_kludge
503 Type: kernel internal
507 This field is obsolete.
509 Field name: heap_end_ptr
510 Type: write (obligatory)
514 Set this field to the offset (from the beginning of the real-mode
515 code) of the end of the setup stack/heap, minus 0x0200.
517 Field name: ext_loader_ver
518 Type: write (optional)
522 This field is used as an extension of the version number in the
523 type_of_loader field. The total version number is considered to be
524 (type_of_loader & 0x0f) + (ext_loader_ver << 4).
526 The use of this field is boot loader specific. If not written, it
529 Kernels prior to 2.6.31 did not recognize this field, but it is safe
530 to write for protocol version 2.02 or higher.
532 Field name: ext_loader_type
533 Type: write (obligatory if (type_of_loader & 0xf0) == 0xe0)
537 This field is used as an extension of the type number in
538 type_of_loader field. If the type in type_of_loader is 0xE, then
539 the actual type is (ext_loader_type + 0x10).
541 This field is ignored if the type in type_of_loader is not 0xE.
543 Kernels prior to 2.6.31 did not recognize this field, but it is safe
544 to write for protocol version 2.02 or higher.
546 Field name: cmd_line_ptr
547 Type: write (obligatory)
551 Set this field to the linear address of the kernel command line.
552 The kernel command line can be located anywhere between the end of
553 the setup heap and 0xA0000; it does not have to be located in the
554 same 64K segment as the real-mode code itself.
556 Fill in this field even if your boot loader does not support a
557 command line, in which case you can point this to an empty string
558 (or better yet, to the string "auto".) If this field is left at
559 zero, the kernel will assume that your boot loader does not support
562 Field name: initrd_addr_max
567 The maximum address that may be occupied by the initial
568 ramdisk/ramfs contents. For boot protocols 2.02 or earlier, this
569 field is not present, and the maximum address is 0x37FFFFFF. (This
570 address is defined as the address of the highest safe byte, so if
571 your ramdisk is exactly 131072 bytes long and this field is
572 0x37FFFFFF, you can start your ramdisk at 0x37FE0000.)
574 Field name: kernel_alignment
575 Type: read/modify (reloc)
577 Protocol: 2.05+ (read), 2.10+ (modify)
579 Alignment unit required by the kernel (if relocatable_kernel is
580 true.) A relocatable kernel that is loaded at an alignment
581 incompatible with the value in this field will be realigned during
582 kernel initialization.
584 Starting with protocol version 2.10, this reflects the kernel
585 alignment preferred for optimal performance; it is possible for the
586 loader to modify this field to permit a lesser alignment. See the
587 min_alignment and pref_address field below.
589 Field name: relocatable_kernel
594 If this field is nonzero, the protected-mode part of the kernel can
595 be loaded at any address that satisfies the kernel_alignment field.
596 After loading, the boot loader must set the code32_start field to
597 point to the loaded code, or to a boot loader hook.
599 Field name: min_alignment
604 This field, if nonzero, indicates as a power of two the minimum
605 alignment required, as opposed to preferred, by the kernel to boot.
606 If a boot loader makes use of this field, it should update the
607 kernel_alignment field with the alignment unit desired; typically:
609 kernel_alignment = 1 << min_alignment
611 There may be a considerable performance cost with an excessively
612 misaligned kernel. Therefore, a loader should typically try each
613 power-of-two alignment from kernel_alignment down to this alignment.
615 Field name: xloadflags
620 This field is a bitmask.
622 Bit 0 (read): XLF_KERNEL_64
623 - If 1, this kernel has the legacy 64-bit entry point at 0x200.
625 Bit 1 (read): XLF_CAN_BE_LOADED_ABOVE_4G
626 - If 1, kernel/boot_params/cmdline/ramdisk can be above 4G.
628 Bit 2 (read): XLF_EFI_HANDOVER_32
629 - If 1, the kernel supports the 32-bit EFI handoff entry point
630 given at handover_offset.
632 Bit 3 (read): XLF_EFI_HANDOVER_64
633 - If 1, the kernel supports the 64-bit EFI handoff entry point
634 given at handover_offset + 0x200.
636 Bit 4 (read): XLF_EFI_KEXEC
637 - If 1, the kernel supports kexec EFI boot with EFI runtime support.
639 Field name: cmdline_size
644 The maximum size of the command line without the terminating
645 zero. This means that the command line can contain at most
646 cmdline_size characters. With protocol version 2.05 and earlier, the
647 maximum size was 255.
649 Field name: hardware_subarch
650 Type: write (optional, defaults to x86/PC)
654 In a paravirtualized environment the hardware low level architectural
655 pieces such as interrupt handling, page table handling, and
656 accessing process control registers needs to be done differently.
658 This field allows the bootloader to inform the kernel we are in one
659 one of those environments.
661 0x00000000 The default x86/PC environment
664 0x00000003 Moorestown MID
665 0x00000004 CE4100 TV Platform
667 Field name: hardware_subarch_data
668 Type: write (subarch-dependent)
672 A pointer to data that is specific to hardware subarch
673 This field is currently unused for the default x86/PC environment,
676 Field name: payload_offset
681 If non-zero then this field contains the offset from the beginning
682 of the protected-mode code to the payload.
684 The payload may be compressed. The format of both the compressed and
685 uncompressed data should be determined using the standard magic
686 numbers. The currently supported compression formats are gzip
687 (magic numbers 1F 8B or 1F 9E), bzip2 (magic number 42 5A), LZMA
688 (magic number 5D 00), XZ (magic number FD 37), and LZ4 (magic number
689 02 21). The uncompressed payload is currently always ELF (magic
692 Field name: payload_length
697 The length of the payload.
699 Field name: setup_data
700 Type: write (special)
704 The 64-bit physical pointer to NULL terminated single linked list of
705 struct setup_data. This is used to define a more extensible boot
706 parameters passing mechanism. The definition of struct setup_data is
716 Where, the next is a 64-bit physical pointer to the next node of
717 linked list, the next field of the last node is 0; the type is used
718 to identify the contents of data; the len is the length of data
719 field; the data holds the real payload.
721 This list may be modified at a number of points during the bootup
722 process. Therefore, when modifying this list one should always make
723 sure to consider the case where the linked list already contains
726 Field name: pref_address
731 This field, if nonzero, represents a preferred load address for the
732 kernel. A relocating bootloader should attempt to load at this
735 A non-relocatable kernel will unconditionally move itself and to run
738 Field name: init_size
742 This field indicates the amount of linear contiguous memory starting
743 at the kernel runtime start address that the kernel needs before it
744 is capable of examining its memory map. This is not the same thing
745 as the total amount of memory the kernel needs to boot, but it can
746 be used by a relocating boot loader to help select a safe load
747 address for the kernel.
749 The kernel runtime start address is determined by the following algorithm:
751 if (relocatable_kernel)
752 runtime_start = align_up(load_address, kernel_alignment)
754 runtime_start = pref_address
756 Field name: handover_offset
760 This field is the offset from the beginning of the kernel image to
761 the EFI handover protocol entry point. Boot loaders using the EFI
762 handover protocol to boot the kernel should jump to this offset.
764 See EFI HANDOVER PROTOCOL below for more details.
766 Field name: acpi_rsdp_addr
771 This field can be set by the boot loader to tell the kernel the
772 physical address of the ACPI RSDP table.
774 A value of 0 indicates the kernel should fall back to the standard
775 methods to locate the RSDP.
778 **** THE IMAGE CHECKSUM
780 From boot protocol version 2.08 onwards the CRC-32 is calculated over
781 the entire file using the characteristic polynomial 0x04C11DB7 and an
782 initial remainder of 0xffffffff. The checksum is appended to the
783 file; therefore the CRC of the file up to the limit specified in the
784 syssize field of the header is always 0.
787 **** THE KERNEL COMMAND LINE
789 The kernel command line has become an important way for the boot
790 loader to communicate with the kernel. Some of its options are also
791 relevant to the boot loader itself, see "special command line options"
794 The kernel command line is a null-terminated string. The maximum
795 length can be retrieved from the field cmdline_size. Before protocol
796 version 2.06, the maximum was 255 characters. A string that is too
797 long will be automatically truncated by the kernel.
799 If the boot protocol version is 2.02 or later, the address of the
800 kernel command line is given by the header field cmd_line_ptr (see
801 above.) This address can be anywhere between the end of the setup
804 If the protocol version is *not* 2.02 or higher, the kernel
805 command line is entered using the following protocol:
807 At offset 0x0020 (word), "cmd_line_magic", enter the magic
810 At offset 0x0022 (word), "cmd_line_offset", enter the offset
811 of the kernel command line (relative to the start of the
814 The kernel command line *must* be within the memory region
815 covered by setup_move_size, so you may need to adjust this
819 **** MEMORY LAYOUT OF THE REAL-MODE CODE
821 The real-mode code requires a stack/heap to be set up, as well as
822 memory allocated for the kernel command line. This needs to be done
823 in the real-mode accessible memory in bottom megabyte.
825 It should be noted that modern machines often have a sizable Extended
826 BIOS Data Area (EBDA). As a result, it is advisable to use as little
827 of the low megabyte as possible.
829 Unfortunately, under the following circumstances the 0x90000 memory
830 segment has to be used:
832 - When loading a zImage kernel ((loadflags & 0x01) == 0).
833 - When loading a 2.01 or earlier boot protocol kernel.
835 -> For the 2.00 and 2.01 boot protocols, the real-mode code
836 can be loaded at another address, but it is internally
837 relocated to 0x90000. For the "old" protocol, the
838 real-mode code must be loaded at 0x90000.
840 When loading at 0x90000, avoid using memory above 0x9a000.
842 For boot protocol 2.02 or higher, the command line does not have to be
843 located in the same 64K segment as the real-mode setup code; it is
844 thus permitted to give the stack/heap the full 64K segment and locate
845 the command line above it.
847 The kernel command line should not be located below the real-mode
848 code, nor should it be located in high memory.
851 **** SAMPLE BOOT CONFIGURATION
853 As a sample configuration, assume the following layout of the real
856 When loading below 0x90000, use the entire segment:
858 0x0000-0x7fff Real mode kernel
859 0x8000-0xdfff Stack and heap
860 0xe000-0xffff Kernel command line
862 When loading at 0x90000 OR the protocol version is 2.01 or earlier:
864 0x0000-0x7fff Real mode kernel
865 0x8000-0x97ff Stack and heap
866 0x9800-0x9fff Kernel command line
868 Such a boot loader should enter the following fields in the header:
870 unsigned long base_ptr; /* base address for real-mode segment */
872 if ( setup_sects == 0 ) {
876 if ( protocol >= 0x0200 ) {
877 type_of_loader = <type code>;
878 if ( loading_initrd ) {
879 ramdisk_image = <initrd_address>;
880 ramdisk_size = <initrd_size>;
883 if ( protocol >= 0x0202 && loadflags & 0x01 )
888 if ( protocol >= 0x0201 ) {
889 heap_end_ptr = heap_end - 0x200;
890 loadflags |= 0x80; /* CAN_USE_HEAP */
893 if ( protocol >= 0x0202 ) {
894 cmd_line_ptr = base_ptr + heap_end;
895 strcpy(cmd_line_ptr, cmdline);
897 cmd_line_magic = 0xA33F;
898 cmd_line_offset = heap_end;
899 setup_move_size = heap_end + strlen(cmdline)+1;
900 strcpy(base_ptr+cmd_line_offset, cmdline);
903 /* Very old kernel */
907 cmd_line_magic = 0xA33F;
908 cmd_line_offset = heap_end;
910 /* A very old kernel MUST have its real-mode code
913 if ( base_ptr != 0x90000 ) {
914 /* Copy the real-mode kernel */
915 memcpy(0x90000, base_ptr, (setup_sects+1)*512);
916 base_ptr = 0x90000; /* Relocated */
919 strcpy(0x90000+cmd_line_offset, cmdline);
921 /* It is recommended to clear memory up to the 32K mark */
922 memset(0x90000 + (setup_sects+1)*512, 0,
923 (64-(setup_sects+1))*512);
927 **** LOADING THE REST OF THE KERNEL
929 The 32-bit (non-real-mode) kernel starts at offset (setup_sects+1)*512
930 in the kernel file (again, if setup_sects == 0 the real value is 4.)
931 It should be loaded at address 0x10000 for Image/zImage kernels and
932 0x100000 for bzImage kernels.
934 The kernel is a bzImage kernel if the protocol >= 2.00 and the 0x01
935 bit (LOAD_HIGH) in the loadflags field is set:
937 is_bzImage = (protocol >= 0x0200) && (loadflags & 0x01);
938 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.
946 **** SPECIAL COMMAND LINE OPTIONS
948 If the command line provided by the boot loader is entered by the
949 user, the user may expect the following command line options to work.
950 They should normally not be deleted from the kernel command line even
951 though not all of them are actually meaningful to the kernel. Boot
952 loader authors who need additional command line options for the boot
953 loader itself should get them registered in
954 Documentation/admin-guide/kernel-parameters.rst to make sure they will not
955 conflict with actual kernel options now or in the future.
958 <mode> here is either an integer (in C notation, either
959 decimal, octal, or hexadecimal) or one of the strings
960 "normal" (meaning 0xFFFF), "ext" (meaning 0xFFFE) or "ask"
961 (meaning 0xFFFD). This value should be entered into the
962 vid_mode field, as it is used by the kernel before the command
966 <size> is an integer in C notation optionally followed by
967 (case insensitive) K, M, G, T, P or E (meaning << 10, << 20,
968 << 30, << 40, << 50 or << 60). This specifies the end of
969 memory to the kernel. This affects the possible placement of
970 an initrd, since an initrd should be placed near end of
971 memory. Note that this is an option to *both* the kernel and
975 An initrd should be loaded. The meaning of <file> is
976 obviously bootloader-dependent, and some boot loaders
977 (e.g. LILO) do not have such a command.
979 In addition, some boot loaders add the following options to the
980 user-specified command line:
983 The boot image which was loaded. Again, the meaning of <file>
984 is obviously bootloader-dependent.
987 The kernel was booted without explicit user intervention.
989 If these options are added by the boot loader, it is highly
990 recommended that they are located *first*, before the user-specified
991 or configuration-specified command line. Otherwise, "init=/bin/sh"
992 gets confused by the "auto" option.
995 **** RUNNING THE KERNEL
997 The kernel is started by jumping to the kernel entry point, which is
998 located at *segment* offset 0x20 from the start of the real mode
999 kernel. This means that if you loaded your real-mode kernel code at
1000 0x90000, the kernel entry point is 9020:0000.
1002 At entry, ds = es = ss should point to the start of the real-mode
1003 kernel code (0x9000 if the code is loaded at 0x90000), sp should be
1004 set up properly, normally pointing to the top of the heap, and
1005 interrupts should be disabled. Furthermore, to guard against bugs in
1006 the kernel, it is recommended that the boot loader sets fs = gs = ds =
1009 In our example from above, we would do:
1011 /* Note: in the case of the "old" kernel protocol, base_ptr must
1012 be == 0x90000 at this point; see the previous sample code */
1014 seg = base_ptr >> 4;
1016 cli(); /* Enter with interrupts disabled! */
1018 /* Set up the real-mode kernel stack */
1022 _DS = _ES = _FS = _GS = seg;
1023 jmp_far(seg+0x20, 0); /* Run the kernel */
1025 If your boot sector accesses a floppy drive, it is recommended to
1026 switch off the floppy motor before running the kernel, since the
1027 kernel boot leaves interrupts off and thus the motor will not be
1028 switched off, especially if the loaded kernel has the floppy driver as
1029 a demand-loaded module!
1032 **** ADVANCED BOOT LOADER HOOKS
1034 If the boot loader runs in a particularly hostile environment (such as
1035 LOADLIN, which runs under DOS) it may be impossible to follow the
1036 standard memory location requirements. Such a boot loader may use the
1037 following hooks that, if set, are invoked by the kernel at the
1038 appropriate time. The use of these hooks should probably be
1039 considered an absolutely last resort!
1041 IMPORTANT: All the hooks are required to preserve %esp, %ebp, %esi and
1042 %edi across invocation.
1045 A 16-bit real mode far subroutine invoked immediately before
1046 entering protected mode. The default routine disables NMI, so
1047 your routine should probably do so, too.
1050 A 32-bit flat-mode routine *jumped* to immediately after the
1051 transition to protected mode, but before the kernel is
1052 uncompressed. No segments, except CS, are guaranteed to be
1053 set up (current kernels do, but older ones do not); you should
1054 set them up to BOOT_DS (0x18) yourself.
1056 After completing your hook, you should jump to the address
1057 that was in this field before your boot loader overwrote it
1058 (relocated, if appropriate.)
1061 **** 32-bit BOOT PROTOCOL
1063 For machine with some new BIOS other than legacy BIOS, such as EFI,
1064 LinuxBIOS, etc, and kexec, the 16-bit real mode setup code in kernel
1065 based on legacy BIOS can not be used, so a 32-bit boot protocol needs
1068 In 32-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 should be allocated and initialized to all zero. Then the setup header
1072 from offset 0x01f1 of kernel image on should be loaded into struct
1073 boot_params and examined. The end of setup header can be calculated as
1076 0x0202 + byte value at offset 0x0201
1078 In addition to read/modify/write the setup header of the struct
1079 boot_params as that of 16-bit boot protocol, the boot loader should
1080 also fill the additional fields of the struct boot_params as that
1081 described in zero-page.txt.
1083 After setting up the struct boot_params, the boot loader can load the
1084 32/64-bit kernel in the same way as that of 16-bit boot protocol.
1086 In 32-bit boot protocol, the kernel is started by jumping to the
1087 32-bit kernel entry point, which is the start address of loaded
1090 At entry, the CPU must be in 32-bit protected mode with paging
1091 disabled; a GDT must be loaded with the descriptors for selectors
1092 __BOOT_CS(0x10) and __BOOT_DS(0x18); both descriptors must be 4G flat
1093 segment; __BOOT_CS must have execute/read permission, and __BOOT_DS
1094 must have read/write permission; CS must be __BOOT_CS and DS, ES, SS
1095 must be __BOOT_DS; interrupt must be disabled; %esi must hold the base
1096 address of the struct boot_params; %ebp, %edi and %ebx must be zero.
1098 **** 64-bit BOOT PROTOCOL
1100 For machine with 64bit cpus and 64bit kernel, we could use 64bit bootloader
1101 and we need a 64-bit boot protocol.
1103 In 64-bit boot protocol, the first step in loading a Linux kernel
1104 should be to setup the boot parameters (struct boot_params,
1105 traditionally known as "zero page"). The memory for struct boot_params
1106 could be allocated anywhere (even above 4G) and initialized to all zero.
1107 Then, the setup header at offset 0x01f1 of kernel image on should be
1108 loaded into struct boot_params and examined. The end of setup header
1109 can be calculated as follows:
1111 0x0202 + byte value at offset 0x0201
1113 In addition to read/modify/write the setup header of the struct
1114 boot_params as that of 16-bit boot protocol, the boot loader should
1115 also fill the additional fields of the struct boot_params as described
1118 After setting up the struct boot_params, the boot loader can load
1119 64-bit kernel in the same way as that of 16-bit boot protocol, but
1120 kernel could be loaded above 4G.
1122 In 64-bit boot protocol, the kernel is started by jumping to the
1123 64-bit kernel entry point, which is the start address of loaded
1124 64-bit kernel plus 0x200.
1126 At entry, the CPU must be in 64-bit mode with paging enabled.
1127 The range with setup_header.init_size from start address of loaded
1128 kernel and zero page and command line buffer get ident mapping;
1129 a GDT must be loaded with the descriptors for selectors
1130 __BOOT_CS(0x10) and __BOOT_DS(0x18); both descriptors must be 4G flat
1131 segment; __BOOT_CS must have execute/read permission, and __BOOT_DS
1132 must have read/write permission; CS must be __BOOT_CS and DS, ES, SS
1133 must be __BOOT_DS; interrupt must be disabled; %rsi must hold the base
1134 address of the struct boot_params.
1136 **** EFI HANDOVER PROTOCOL
1138 This protocol allows boot loaders to defer initialisation to the EFI
1139 boot stub. The boot loader is required to load the kernel/initrd(s)
1140 from the boot media and jump to the EFI handover protocol entry point
1141 which is hdr->handover_offset bytes from the beginning of
1144 The function prototype for the handover entry point looks like this,
1146 efi_main(void *handle, efi_system_table_t *table, struct boot_params *bp)
1148 'handle' is the EFI image handle passed to the boot loader by the EFI
1149 firmware, 'table' is the EFI system table - these are the first two
1150 arguments of the "handoff state" as described in section 2.3 of the
1151 UEFI specification. 'bp' is the boot loader-allocated boot params.
1153 The boot loader *must* fill out the following fields in bp,
1157 o hdr.ramdisk_image (if applicable)
1158 o hdr.ramdisk_size (if applicable)
1160 All other fields should be zero.