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
62 The traditional memory map for the kernel loader, used for Image or
63 zImage kernels, typically looks like:
66 0A0000 +------------------------+
67 | Reserved for BIOS | Do not use. Reserved for BIOS EBDA.
68 09A000 +------------------------+
70 | Stack/heap | For use by the kernel real-mode code.
71 098000 +------------------------+
72 | Kernel setup | The kernel real-mode code.
73 090200 +------------------------+
74 | Kernel boot sector | The kernel legacy boot sector.
75 090000 +------------------------+
76 | Protected-mode kernel | The bulk of the kernel image.
77 010000 +------------------------+
78 | Boot loader | <- Boot sector entry point 0000:7C00
79 001000 +------------------------+
80 | Reserved for MBR/BIOS |
81 000800 +------------------------+
82 | Typically used by MBR |
83 000600 +------------------------+
85 000000 +------------------------+
88 When using bzImage, the protected-mode kernel was relocated to
89 0x100000 ("high memory"), and the kernel real-mode block (boot sector,
90 setup, and stack/heap) was made relocatable to any address between
91 0x10000 and end of low memory. Unfortunately, in protocols 2.00 and
92 2.01 the 0x90000+ memory range is still used internally by the kernel;
93 the 2.02 protocol resolves that problem.
95 It is desirable to keep the "memory ceiling" -- the highest point in
96 low memory touched by the boot loader -- as low as possible, since
97 some newer BIOSes have begun to allocate some rather large amounts of
98 memory, called the Extended BIOS Data Area, near the top of low
99 memory. The boot loader should use the "INT 12h" BIOS call to verify
100 how much low memory is available.
102 Unfortunately, if INT 12h reports that the amount of memory is too
103 low, there is usually nothing the boot loader can do but to report an
104 error to the user. The boot loader should therefore be designed to
105 take up as little space in low memory as it reasonably can. For
106 zImage or old bzImage kernels, which need data written into the
107 0x90000 segment, the boot loader should make sure not to use memory
108 above the 0x9A000 point; too many BIOSes will break above that point.
110 For a modern bzImage kernel with boot protocol version >= 2.02, a
111 memory layout like the following is suggested:
114 | Protected-mode kernel |
115 100000 +------------------------+
117 0A0000 +------------------------+
118 | Reserved for BIOS | Leave as much as possible unused
120 | Command line | (Can also be below the X+10000 mark)
121 X+10000 +------------------------+
122 | Stack/heap | For use by the kernel real-mode code.
123 X+08000 +------------------------+
124 | Kernel setup | The kernel real-mode code.
125 | Kernel boot sector | The kernel legacy boot sector.
126 X +------------------------+
127 | Boot loader | <- Boot sector entry point 0000:7C00
128 001000 +------------------------+
129 | Reserved for MBR/BIOS |
130 000800 +------------------------+
131 | Typically used by MBR |
132 000600 +------------------------+
134 000000 +------------------------+
136 ... where the address X is as low as the design of the boot loader
140 **** THE REAL-MODE KERNEL HEADER
142 In the following text, and anywhere in the kernel boot sequence, "a
143 sector" refers to 512 bytes. It is independent of the actual sector
144 size of the underlying medium.
146 The first step in loading a Linux kernel should be to load the
147 real-mode code (boot sector and setup code) and then examine the
148 following header at offset 0x01f1. The real-mode code can total up to
149 32K, although the boot loader may choose to load only the first two
150 sectors (1K) and then examine the bootup sector size.
152 The header looks like:
154 Offset Proto Name Meaning
157 01F1/1 ALL(1 setup_sects The size of the setup in sectors
158 01F2/2 ALL root_flags If set, the root is mounted readonly
159 01F4/4 2.04+(2 syssize The size of the 32-bit code in 16-byte paras
160 01F8/2 ALL ram_size DO NOT USE - for bootsect.S use only
161 01FA/2 ALL vid_mode Video mode control
162 01FC/2 ALL root_dev Default root device number
163 01FE/2 ALL boot_flag 0xAA55 magic number
164 0200/2 2.00+ jump Jump instruction
165 0202/4 2.00+ header Magic signature "HdrS"
166 0206/2 2.00+ version Boot protocol version supported
167 0208/4 2.00+ realmode_swtch Boot loader hook (see below)
168 020C/2 2.00+ start_sys_seg The load-low segment (0x1000) (obsolete)
169 020E/2 2.00+ kernel_version Pointer to kernel version string
170 0210/1 2.00+ type_of_loader Boot loader identifier
171 0211/1 2.00+ loadflags Boot protocol option flags
172 0212/2 2.00+ setup_move_size Move to high memory size (used with hooks)
173 0214/4 2.00+ code32_start Boot loader hook (see below)
174 0218/4 2.00+ ramdisk_image initrd load address (set by boot loader)
175 021C/4 2.00+ ramdisk_size initrd size (set by boot loader)
176 0220/4 2.00+ bootsect_kludge DO NOT USE - for bootsect.S use only
177 0224/2 2.01+ heap_end_ptr Free memory after setup end
178 0226/1 2.02+(3 ext_loader_ver Extended boot loader version
179 0227/1 2.02+(3 ext_loader_type Extended boot loader ID
180 0228/4 2.02+ cmd_line_ptr 32-bit pointer to the kernel command line
181 022C/4 2.03+ ramdisk_max Highest legal initrd address
182 0230/4 2.05+ kernel_alignment Physical addr alignment required for kernel
183 0234/1 2.05+ relocatable_kernel Whether kernel is relocatable or not
184 0235/1 2.10+ min_alignment Minimum alignment, as a power of two
185 0236/2 N/A pad3 Unused
186 0238/4 2.06+ cmdline_size Maximum size of the kernel command line
187 023C/4 2.07+ hardware_subarch Hardware subarchitecture
188 0240/8 2.07+ hardware_subarch_data Subarchitecture-specific data
189 0248/4 2.08+ payload_offset Offset of kernel payload
190 024C/4 2.08+ payload_length Length of kernel payload
191 0250/8 2.09+ setup_data 64-bit physical pointer to linked list
193 0258/8 2.10+ pref_address Preferred loading address
194 0260/4 2.10+ init_size Linear memory required during initialization
195 0264/4 2.11+ handover_offset Offset of handover entry point
197 (1) For backwards compatibility, if the setup_sects field contains 0, the
200 (2) For boot protocol prior to 2.04, the upper two bytes of the syssize
201 field are unusable, which means the size of a bzImage kernel
202 cannot be determined.
204 (3) Ignored, but safe to set, for boot protocols 2.02-2.09.
206 If the "HdrS" (0x53726448) magic number is not found at offset 0x202,
207 the boot protocol version is "old". Loading an old kernel, the
208 following parameters should be assumed:
212 Real-mode kernel must be located at 0x90000.
214 Otherwise, the "version" field contains the protocol version,
215 e.g. protocol version 2.01 will contain 0x0201 in this field. When
216 setting fields in the header, you must make sure only to set fields
217 supported by the protocol version in use.
220 **** DETAILS OF HEADER FIELDS
222 For each field, some are information from the kernel to the bootloader
223 ("read"), some are expected to be filled out by the bootloader
224 ("write"), and some are expected to be read and modified by the
225 bootloader ("modify").
227 All general purpose boot loaders should write the fields marked
228 (obligatory). Boot loaders who want to load the kernel at a
229 nonstandard address should fill in the fields marked (reloc); other
230 boot loaders can ignore those fields.
232 The byte order of all fields is littleendian (this is x86, after all.)
234 Field name: setup_sects
239 The size of the setup code in 512-byte sectors. If this field is
240 0, the real value is 4. The real-mode code consists of the boot
241 sector (always one 512-byte sector) plus the setup code.
243 Field name: root_flags
244 Type: modify (optional)
248 If this field is nonzero, the root defaults to readonly. The use of
249 this field is deprecated; use the "ro" or "rw" options on the
250 command line instead.
254 Offset/size: 0x1f4/4 (protocol 2.04+) 0x1f4/2 (protocol ALL)
257 The size of the protected-mode code in units of 16-byte paragraphs.
258 For protocol versions older than 2.04 this field is only two bytes
259 wide, and therefore cannot be trusted for the size of a kernel if
260 the LOAD_HIGH flag is set.
263 Type: kernel internal
267 This field is obsolete.
270 Type: modify (obligatory)
273 Please see the section on SPECIAL COMMAND LINE OPTIONS.
276 Type: modify (optional)
280 The default root device device number. The use of this field is
281 deprecated, use the "root=" option on the command line instead.
283 Field name: boot_flag
288 Contains 0xAA55. This is the closest thing old Linux kernels have
296 Contains an x86 jump instruction, 0xEB followed by a signed offset
297 relative to byte 0x202. This can be used to determine the size of
305 Contains the magic number "HdrS" (0x53726448).
312 Contains the boot protocol version, in (major << 8)+minor format,
313 e.g. 0x0204 for version 2.04, and 0x0a11 for a hypothetical version
316 Field name: realmode_swtch
317 Type: modify (optional)
321 Boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
323 Field name: start_sys_seg
328 The load low segment (0x1000). Obsolete.
330 Field name: kernel_version
335 If set to a nonzero value, contains a pointer to a NUL-terminated
336 human-readable kernel version number string, less 0x200. This can
337 be used to display the kernel version to the user. This value
338 should be less than (0x200*setup_sects).
340 For example, if this value is set to 0x1c00, the kernel version
341 number string can be found at offset 0x1e00 in the kernel file.
342 This is a valid value if and only if the "setup_sects" field
343 contains the value 15 or higher, as:
345 0x1c00 < 15*0x200 (= 0x1e00) but
346 0x1c00 >= 14*0x200 (= 0x1c00)
348 0x1c00 >> 9 = 14, so the minimum value for setup_secs is 15.
350 Field name: type_of_loader
351 Type: write (obligatory)
355 If your boot loader has an assigned id (see table below), enter
356 0xTV here, where T is an identifier for the boot loader and V is
357 a version number. Otherwise, enter 0xFF here.
359 For boot loader IDs above T = 0xD, write T = 0xE to this field and
360 write the extended ID minus 0x10 to the ext_loader_type field.
361 Similarly, the ext_loader_ver field can be used to provide more than
362 four bits for the bootloader version.
364 For example, for T = 0x15, V = 0x234, write:
366 type_of_loader <- 0xE4
367 ext_loader_type <- 0x05
368 ext_loader_ver <- 0x23
370 Assigned boot loader ids (hexadecimal):
372 0 LILO (0x00 reserved for pre-2.00 bootloader)
374 2 bootsect-loader (0x20, all other values reserved)
376 4 Etherboot/gPXE/iPXE
383 C Arcturus Networks uCbootloader
385 E Extended (see ext_loader_type)
386 F Special (0xFF = undefined)
388 11 Minimal Linux Bootloader <http://sebastian-plotz.blogspot.de>
390 Please contact <hpa@zytor.com> if you need a bootloader ID
393 Field name: loadflags
394 Type: modify (obligatory)
398 This field is a bitmask.
400 Bit 0 (read): LOADED_HIGH
401 - If 0, the protected-mode code is loaded at 0x10000.
402 - If 1, the protected-mode code is loaded at 0x100000.
404 Bit 5 (write): QUIET_FLAG
405 - If 0, print early messages.
406 - If 1, suppress early messages.
407 This requests to the kernel (decompressor and early
408 kernel) to not write early messages that require
409 accessing the display hardware directly.
411 Bit 6 (write): KEEP_SEGMENTS
413 - If 0, reload the segment registers in the 32bit entry point.
414 - If 1, do not reload the segment registers in the 32bit entry point.
415 Assume that %cs %ds %ss %es are all set to flat segments with
416 a base of 0 (or the equivalent for their environment).
418 Bit 7 (write): CAN_USE_HEAP
419 Set this bit to 1 to indicate that the value entered in the
420 heap_end_ptr is valid. If this field is clear, some setup code
421 functionality will be disabled.
423 Field name: setup_move_size
424 Type: modify (obligatory)
428 When using protocol 2.00 or 2.01, if the real mode kernel is not
429 loaded at 0x90000, it gets moved there later in the loading
430 sequence. Fill in this field if you want additional data (such as
431 the kernel command line) moved in addition to the real-mode kernel
434 The unit is bytes starting with the beginning of the boot sector.
436 This field is can be ignored when the protocol is 2.02 or higher, or
437 if the real-mode code is loaded at 0x90000.
439 Field name: code32_start
440 Type: modify (optional, reloc)
444 The address to jump to in protected mode. This defaults to the load
445 address of the kernel, and can be used by the boot loader to
446 determine the proper load address.
448 This field can be modified for two purposes:
450 1. as a boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
452 2. if a bootloader which does not install a hook loads a
453 relocatable kernel at a nonstandard address it will have to modify
454 this field to point to the load address.
456 Field name: ramdisk_image
457 Type: write (obligatory)
461 The 32-bit linear address of the initial ramdisk or ramfs. Leave at
462 zero if there is no initial ramdisk/ramfs.
464 Field name: ramdisk_size
465 Type: write (obligatory)
469 Size of the initial ramdisk or ramfs. Leave at zero if there is no
470 initial ramdisk/ramfs.
472 Field name: bootsect_kludge
473 Type: kernel internal
477 This field is obsolete.
479 Field name: heap_end_ptr
480 Type: write (obligatory)
484 Set this field to the offset (from the beginning of the real-mode
485 code) of the end of the setup stack/heap, minus 0x0200.
487 Field name: ext_loader_ver
488 Type: write (optional)
492 This field is used as an extension of the version number in the
493 type_of_loader field. The total version number is considered to be
494 (type_of_loader & 0x0f) + (ext_loader_ver << 4).
496 The use of this field is boot loader specific. If not written, it
499 Kernels prior to 2.6.31 did not recognize this field, but it is safe
500 to write for protocol version 2.02 or higher.
502 Field name: ext_loader_type
503 Type: write (obligatory if (type_of_loader & 0xf0) == 0xe0)
507 This field is used as an extension of the type number in
508 type_of_loader field. If the type in type_of_loader is 0xE, then
509 the actual type is (ext_loader_type + 0x10).
511 This field is ignored if the type in type_of_loader is not 0xE.
513 Kernels prior to 2.6.31 did not recognize this field, but it is safe
514 to write for protocol version 2.02 or higher.
516 Field name: cmd_line_ptr
517 Type: write (obligatory)
521 Set this field to the linear address of the kernel command line.
522 The kernel command line can be located anywhere between the end of
523 the setup heap and 0xA0000; it does not have to be located in the
524 same 64K segment as the real-mode code itself.
526 Fill in this field even if your boot loader does not support a
527 command line, in which case you can point this to an empty string
528 (or better yet, to the string "auto".) If this field is left at
529 zero, the kernel will assume that your boot loader does not support
532 Field name: ramdisk_max
537 The maximum address that may be occupied by the initial
538 ramdisk/ramfs contents. For boot protocols 2.02 or earlier, this
539 field is not present, and the maximum address is 0x37FFFFFF. (This
540 address is defined as the address of the highest safe byte, so if
541 your ramdisk is exactly 131072 bytes long and this field is
542 0x37FFFFFF, you can start your ramdisk at 0x37FE0000.)
544 Field name: kernel_alignment
545 Type: read/modify (reloc)
547 Protocol: 2.05+ (read), 2.10+ (modify)
549 Alignment unit required by the kernel (if relocatable_kernel is
550 true.) A relocatable kernel that is loaded at an alignment
551 incompatible with the value in this field will be realigned during
552 kernel initialization.
554 Starting with protocol version 2.10, this reflects the kernel
555 alignment preferred for optimal performance; it is possible for the
556 loader to modify this field to permit a lesser alignment. See the
557 min_alignment and pref_address field below.
559 Field name: relocatable_kernel
564 If this field is nonzero, the protected-mode part of the kernel can
565 be loaded at any address that satisfies the kernel_alignment field.
566 After loading, the boot loader must set the code32_start field to
567 point to the loaded code, or to a boot loader hook.
569 Field name: min_alignment
574 This field, if nonzero, indicates as a power of two the minimum
575 alignment required, as opposed to preferred, by the kernel to boot.
576 If a boot loader makes use of this field, it should update the
577 kernel_alignment field with the alignment unit desired; typically:
579 kernel_alignment = 1 << min_alignment
581 There may be a considerable performance cost with an excessively
582 misaligned kernel. Therefore, a loader should typically try each
583 power-of-two alignment from kernel_alignment down to this alignment.
585 Field name: cmdline_size
590 The maximum size of the command line without the terminating
591 zero. This means that the command line can contain at most
592 cmdline_size characters. With protocol version 2.05 and earlier, the
593 maximum size was 255.
595 Field name: hardware_subarch
596 Type: write (optional, defaults to x86/PC)
600 In a paravirtualized environment the hardware low level architectural
601 pieces such as interrupt handling, page table handling, and
602 accessing process control registers needs to be done differently.
604 This field allows the bootloader to inform the kernel we are in one
605 one of those environments.
607 0x00000000 The default x86/PC environment
610 0x00000003 Moorestown MID
611 0x00000004 CE4100 TV Platform
613 Field name: hardware_subarch_data
614 Type: write (subarch-dependent)
618 A pointer to data that is specific to hardware subarch
619 This field is currently unused for the default x86/PC environment,
622 Field name: payload_offset
627 If non-zero then this field contains the offset from the beginning
628 of the protected-mode code to the payload.
630 The payload may be compressed. The format of both the compressed and
631 uncompressed data should be determined using the standard magic
632 numbers. The currently supported compression formats are gzip
633 (magic numbers 1F 8B or 1F 9E), bzip2 (magic number 42 5A), LZMA
634 (magic number 5D 00), and XZ (magic number FD 37). The uncompressed
635 payload is currently always ELF (magic number 7F 45 4C 46).
637 Field name: payload_length
642 The length of the payload.
644 Field name: setup_data
645 Type: write (special)
649 The 64-bit physical pointer to NULL terminated single linked list of
650 struct setup_data. This is used to define a more extensible boot
651 parameters passing mechanism. The definition of struct setup_data is
661 Where, the next is a 64-bit physical pointer to the next node of
662 linked list, the next field of the last node is 0; the type is used
663 to identify the contents of data; the len is the length of data
664 field; the data holds the real payload.
666 This list may be modified at a number of points during the bootup
667 process. Therefore, when modifying this list one should always make
668 sure to consider the case where the linked list already contains
671 Field name: pref_address
676 This field, if nonzero, represents a preferred load address for the
677 kernel. A relocating bootloader should attempt to load at this
680 A non-relocatable kernel will unconditionally move itself and to run
683 Field name: init_size
687 This field indicates the amount of linear contiguous memory starting
688 at the kernel runtime start address that the kernel needs before it
689 is capable of examining its memory map. This is not the same thing
690 as the total amount of memory the kernel needs to boot, but it can
691 be used by a relocating boot loader to help select a safe load
692 address for the kernel.
694 The kernel runtime start address is determined by the following algorithm:
696 if (relocatable_kernel)
697 runtime_start = align_up(load_address, kernel_alignment)
699 runtime_start = pref_address
701 Field name: handover_offset
705 This field is the offset from the beginning of the kernel image to
706 the EFI handover protocol entry point. Boot loaders using the EFI
707 handover protocol to boot the kernel should jump to this offset.
709 See EFI HANDOVER PROTOCOL below for more details.
712 **** THE IMAGE CHECKSUM
714 From boot protocol version 2.08 onwards the CRC-32 is calculated over
715 the entire file using the characteristic polynomial 0x04C11DB7 and an
716 initial remainder of 0xffffffff. The checksum is appended to the
717 file; therefore the CRC of the file up to the limit specified in the
718 syssize field of the header is always 0.
721 **** THE KERNEL COMMAND LINE
723 The kernel command line has become an important way for the boot
724 loader to communicate with the kernel. Some of its options are also
725 relevant to the boot loader itself, see "special command line options"
728 The kernel command line is a null-terminated string. The maximum
729 length can be retrieved from the field cmdline_size. Before protocol
730 version 2.06, the maximum was 255 characters. A string that is too
731 long will be automatically truncated by the kernel.
733 If the boot protocol version is 2.02 or later, the address of the
734 kernel command line is given by the header field cmd_line_ptr (see
735 above.) This address can be anywhere between the end of the setup
738 If the protocol version is *not* 2.02 or higher, the kernel
739 command line is entered using the following protocol:
741 At offset 0x0020 (word), "cmd_line_magic", enter the magic
744 At offset 0x0022 (word), "cmd_line_offset", enter the offset
745 of the kernel command line (relative to the start of the
748 The kernel command line *must* be within the memory region
749 covered by setup_move_size, so you may need to adjust this
753 **** MEMORY LAYOUT OF THE REAL-MODE CODE
755 The real-mode code requires a stack/heap to be set up, as well as
756 memory allocated for the kernel command line. This needs to be done
757 in the real-mode accessible memory in bottom megabyte.
759 It should be noted that modern machines often have a sizable Extended
760 BIOS Data Area (EBDA). As a result, it is advisable to use as little
761 of the low megabyte as possible.
763 Unfortunately, under the following circumstances the 0x90000 memory
764 segment has to be used:
766 - When loading a zImage kernel ((loadflags & 0x01) == 0).
767 - When loading a 2.01 or earlier boot protocol kernel.
769 -> For the 2.00 and 2.01 boot protocols, the real-mode code
770 can be loaded at another address, but it is internally
771 relocated to 0x90000. For the "old" protocol, the
772 real-mode code must be loaded at 0x90000.
774 When loading at 0x90000, avoid using memory above 0x9a000.
776 For boot protocol 2.02 or higher, the command line does not have to be
777 located in the same 64K segment as the real-mode setup code; it is
778 thus permitted to give the stack/heap the full 64K segment and locate
779 the command line above it.
781 The kernel command line should not be located below the real-mode
782 code, nor should it be located in high memory.
785 **** SAMPLE BOOT CONFIGURATION
787 As a sample configuration, assume the following layout of the real
790 When loading below 0x90000, use the entire segment:
792 0x0000-0x7fff Real mode kernel
793 0x8000-0xdfff Stack and heap
794 0xe000-0xffff Kernel command line
796 When loading at 0x90000 OR the protocol version is 2.01 or earlier:
798 0x0000-0x7fff Real mode kernel
799 0x8000-0x97ff Stack and heap
800 0x9800-0x9fff Kernel command line
802 Such a boot loader should enter the following fields in the header:
804 unsigned long base_ptr; /* base address for real-mode segment */
806 if ( setup_sects == 0 ) {
810 if ( protocol >= 0x0200 ) {
811 type_of_loader = <type code>;
812 if ( loading_initrd ) {
813 ramdisk_image = <initrd_address>;
814 ramdisk_size = <initrd_size>;
817 if ( protocol >= 0x0202 && loadflags & 0x01 )
822 if ( protocol >= 0x0201 ) {
823 heap_end_ptr = heap_end - 0x200;
824 loadflags |= 0x80; /* CAN_USE_HEAP */
827 if ( protocol >= 0x0202 ) {
828 cmd_line_ptr = base_ptr + heap_end;
829 strcpy(cmd_line_ptr, cmdline);
831 cmd_line_magic = 0xA33F;
832 cmd_line_offset = heap_end;
833 setup_move_size = heap_end + strlen(cmdline)+1;
834 strcpy(base_ptr+cmd_line_offset, cmdline);
837 /* Very old kernel */
841 cmd_line_magic = 0xA33F;
842 cmd_line_offset = heap_end;
844 /* A very old kernel MUST have its real-mode code
847 if ( base_ptr != 0x90000 ) {
848 /* Copy the real-mode kernel */
849 memcpy(0x90000, base_ptr, (setup_sects+1)*512);
850 base_ptr = 0x90000; /* Relocated */
853 strcpy(0x90000+cmd_line_offset, cmdline);
855 /* It is recommended to clear memory up to the 32K mark */
856 memset(0x90000 + (setup_sects+1)*512, 0,
857 (64-(setup_sects+1))*512);
861 **** LOADING THE REST OF THE KERNEL
863 The 32-bit (non-real-mode) kernel starts at offset (setup_sects+1)*512
864 in the kernel file (again, if setup_sects == 0 the real value is 4.)
865 It should be loaded at address 0x10000 for Image/zImage kernels and
866 0x100000 for bzImage kernels.
868 The kernel is a bzImage kernel if the protocol >= 2.00 and the 0x01
869 bit (LOAD_HIGH) in the loadflags field is set:
871 is_bzImage = (protocol >= 0x0200) && (loadflags & 0x01);
872 load_address = is_bzImage ? 0x100000 : 0x10000;
874 Note that Image/zImage kernels can be up to 512K in size, and thus use
875 the entire 0x10000-0x90000 range of memory. This means it is pretty
876 much a requirement for these kernels to load the real-mode part at
877 0x90000. bzImage kernels allow much more flexibility.
880 **** SPECIAL COMMAND LINE OPTIONS
882 If the command line provided by the boot loader is entered by the
883 user, the user may expect the following command line options to work.
884 They should normally not be deleted from the kernel command line even
885 though not all of them are actually meaningful to the kernel. Boot
886 loader authors who need additional command line options for the boot
887 loader itself should get them registered in
888 Documentation/kernel-parameters.txt to make sure they will not
889 conflict with actual kernel options now or in the future.
892 <mode> here is either an integer (in C notation, either
893 decimal, octal, or hexadecimal) or one of the strings
894 "normal" (meaning 0xFFFF), "ext" (meaning 0xFFFE) or "ask"
895 (meaning 0xFFFD). This value should be entered into the
896 vid_mode field, as it is used by the kernel before the command
900 <size> is an integer in C notation optionally followed by
901 (case insensitive) K, M, G, T, P or E (meaning << 10, << 20,
902 << 30, << 40, << 50 or << 60). This specifies the end of
903 memory to the kernel. This affects the possible placement of
904 an initrd, since an initrd should be placed near end of
905 memory. Note that this is an option to *both* the kernel and
909 An initrd should be loaded. The meaning of <file> is
910 obviously bootloader-dependent, and some boot loaders
911 (e.g. LILO) do not have such a command.
913 In addition, some boot loaders add the following options to the
914 user-specified command line:
917 The boot image which was loaded. Again, the meaning of <file>
918 is obviously bootloader-dependent.
921 The kernel was booted without explicit user intervention.
923 If these options are added by the boot loader, it is highly
924 recommended that they are located *first*, before the user-specified
925 or configuration-specified command line. Otherwise, "init=/bin/sh"
926 gets confused by the "auto" option.
929 **** RUNNING THE KERNEL
931 The kernel is started by jumping to the kernel entry point, which is
932 located at *segment* offset 0x20 from the start of the real mode
933 kernel. This means that if you loaded your real-mode kernel code at
934 0x90000, the kernel entry point is 9020:0000.
936 At entry, ds = es = ss should point to the start of the real-mode
937 kernel code (0x9000 if the code is loaded at 0x90000), sp should be
938 set up properly, normally pointing to the top of the heap, and
939 interrupts should be disabled. Furthermore, to guard against bugs in
940 the kernel, it is recommended that the boot loader sets fs = gs = ds =
943 In our example from above, we would do:
945 /* Note: in the case of the "old" kernel protocol, base_ptr must
946 be == 0x90000 at this point; see the previous sample code */
950 cli(); /* Enter with interrupts disabled! */
952 /* Set up the real-mode kernel stack */
956 _DS = _ES = _FS = _GS = seg;
957 jmp_far(seg+0x20, 0); /* Run the kernel */
959 If your boot sector accesses a floppy drive, it is recommended to
960 switch off the floppy motor before running the kernel, since the
961 kernel boot leaves interrupts off and thus the motor will not be
962 switched off, especially if the loaded kernel has the floppy driver as
963 a demand-loaded module!
966 **** ADVANCED BOOT LOADER HOOKS
968 If the boot loader runs in a particularly hostile environment (such as
969 LOADLIN, which runs under DOS) it may be impossible to follow the
970 standard memory location requirements. Such a boot loader may use the
971 following hooks that, if set, are invoked by the kernel at the
972 appropriate time. The use of these hooks should probably be
973 considered an absolutely last resort!
975 IMPORTANT: All the hooks are required to preserve %esp, %ebp, %esi and
976 %edi across invocation.
979 A 16-bit real mode far subroutine invoked immediately before
980 entering protected mode. The default routine disables NMI, so
981 your routine should probably do so, too.
984 A 32-bit flat-mode routine *jumped* to immediately after the
985 transition to protected mode, but before the kernel is
986 uncompressed. No segments, except CS, are guaranteed to be
987 set up (current kernels do, but older ones do not); you should
988 set them up to BOOT_DS (0x18) yourself.
990 After completing your hook, you should jump to the address
991 that was in this field before your boot loader overwrote it
992 (relocated, if appropriate.)
995 **** 32-bit BOOT PROTOCOL
997 For machine with some new BIOS other than legacy BIOS, such as EFI,
998 LinuxBIOS, etc, and kexec, the 16-bit real mode setup code in kernel
999 based on legacy BIOS can not be used, so a 32-bit boot protocol needs
1002 In 32-bit boot protocol, the first step in loading a Linux kernel
1003 should be to setup the boot parameters (struct boot_params,
1004 traditionally known as "zero page"). The memory for struct boot_params
1005 should be allocated and initialized to all zero. Then the setup header
1006 from offset 0x01f1 of kernel image on should be loaded into struct
1007 boot_params and examined. The end of setup header can be calculated as
1010 0x0202 + byte value at offset 0x0201
1012 In addition to read/modify/write the setup header of the struct
1013 boot_params as that of 16-bit boot protocol, the boot loader should
1014 also fill the additional fields of the struct boot_params as that
1015 described in zero-page.txt.
1017 After setting up the struct boot_params, the boot loader can load the
1018 32/64-bit kernel in the same way as that of 16-bit boot protocol.
1020 In 32-bit boot protocol, the kernel is started by jumping to the
1021 32-bit kernel entry point, which is the start address of loaded
1024 At entry, the CPU must be in 32-bit protected mode with paging
1025 disabled; a GDT must be loaded with the descriptors for selectors
1026 __BOOT_CS(0x10) and __BOOT_DS(0x18); both descriptors must be 4G flat
1027 segment; __BOOT_CS must have execute/read permission, and __BOOT_DS
1028 must have read/write permission; CS must be __BOOT_CS and DS, ES, SS
1029 must be __BOOT_DS; interrupt must be disabled; %esi must hold the base
1030 address of the struct boot_params; %ebp, %edi and %ebx must be zero.
1032 **** EFI HANDOVER PROTOCOL
1034 This protocol allows boot loaders to defer initialisation to the EFI
1035 boot stub. The boot loader is required to load the kernel/initrd(s)
1036 from the boot media and jump to the EFI handover protocol entry point
1037 which is hdr->handover_offset bytes from the beginning of
1040 The function prototype for the handover entry point looks like this,
1042 efi_main(void *handle, efi_system_table_t *table, struct boot_params *bp)
1044 'handle' is the EFI image handle passed to the boot loader by the EFI
1045 firmware, 'table' is the EFI system table - these are the first two
1046 arguments of the "handoff state" as described in section 2.3 of the
1047 UEFI specification. 'bp' is the boot loader-allocated boot params.
1049 The boot loader *must* fill out the following fields in bp,
1054 o hdr.ramdisk_image (if applicable)
1055 o hdr.ramdisk_size (if applicable)
1057 All other fields should be zero.