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.
55 The traditional memory map for the kernel loader, used for Image or
56 zImage kernels, typically looks like:
59 0A0000 +------------------------+
60 | Reserved for BIOS | Do not use. Reserved for BIOS EBDA.
61 09A000 +------------------------+
63 | Stack/heap | For use by the kernel real-mode code.
64 098000 +------------------------+
65 | Kernel setup | The kernel real-mode code.
66 090200 +------------------------+
67 | Kernel boot sector | The kernel legacy boot sector.
68 090000 +------------------------+
69 | Protected-mode kernel | The bulk of the kernel image.
70 010000 +------------------------+
71 | Boot loader | <- Boot sector entry point 0000:7C00
72 001000 +------------------------+
73 | Reserved for MBR/BIOS |
74 000800 +------------------------+
75 | Typically used by MBR |
76 000600 +------------------------+
78 000000 +------------------------+
81 When using bzImage, the protected-mode kernel was relocated to
82 0x100000 ("high memory"), and the kernel real-mode block (boot sector,
83 setup, and stack/heap) was made relocatable to any address between
84 0x10000 and end of low memory. Unfortunately, in protocols 2.00 and
85 2.01 the 0x90000+ memory range is still used internally by the kernel;
86 the 2.02 protocol resolves that problem.
88 It is desirable to keep the "memory ceiling" -- the highest point in
89 low memory touched by the boot loader -- as low as possible, since
90 some newer BIOSes have begun to allocate some rather large amounts of
91 memory, called the Extended BIOS Data Area, near the top of low
92 memory. The boot loader should use the "INT 12h" BIOS call to verify
93 how much low memory is available.
95 Unfortunately, if INT 12h reports that the amount of memory is too
96 low, there is usually nothing the boot loader can do but to report an
97 error to the user. The boot loader should therefore be designed to
98 take up as little space in low memory as it reasonably can. For
99 zImage or old bzImage kernels, which need data written into the
100 0x90000 segment, the boot loader should make sure not to use memory
101 above the 0x9A000 point; too many BIOSes will break above that point.
103 For a modern bzImage kernel with boot protocol version >= 2.02, a
104 memory layout like the following is suggested:
107 | Protected-mode kernel |
108 100000 +------------------------+
110 0A0000 +------------------------+
111 | Reserved for BIOS | Leave as much as possible unused
113 | Command line | (Can also be below the X+10000 mark)
114 X+10000 +------------------------+
115 | Stack/heap | For use by the kernel real-mode code.
116 X+08000 +------------------------+
117 | Kernel setup | The kernel real-mode code.
118 | Kernel boot sector | The kernel legacy boot sector.
119 X +------------------------+
120 | Boot loader | <- Boot sector entry point 0000:7C00
121 001000 +------------------------+
122 | Reserved for MBR/BIOS |
123 000800 +------------------------+
124 | Typically used by MBR |
125 000600 +------------------------+
127 000000 +------------------------+
129 ... where the address X is as low as the design of the boot loader
133 **** THE REAL-MODE KERNEL HEADER
135 In the following text, and anywhere in the kernel boot sequence, "a
136 sector" refers to 512 bytes. It is independent of the actual sector
137 size of the underlying medium.
139 The first step in loading a Linux kernel should be to load the
140 real-mode code (boot sector and setup code) and then examine the
141 following header at offset 0x01f1. The real-mode code can total up to
142 32K, although the boot loader may choose to load only the first two
143 sectors (1K) and then examine the bootup sector size.
145 The header looks like:
147 Offset Proto Name Meaning
150 01F1/1 ALL(1 setup_sects The size of the setup in sectors
151 01F2/2 ALL root_flags If set, the root is mounted readonly
152 01F4/4 2.04+(2 syssize The size of the 32-bit code in 16-byte paras
153 01F8/2 ALL ram_size DO NOT USE - for bootsect.S use only
154 01FA/2 ALL vid_mode Video mode control
155 01FC/2 ALL root_dev Default root device number
156 01FE/2 ALL boot_flag 0xAA55 magic number
157 0200/2 2.00+ jump Jump instruction
158 0202/4 2.00+ header Magic signature "HdrS"
159 0206/2 2.00+ version Boot protocol version supported
160 0208/4 2.00+ realmode_swtch Boot loader hook (see below)
161 020C/2 2.00+ start_sys_seg The load-low segment (0x1000) (obsolete)
162 020E/2 2.00+ kernel_version Pointer to kernel version string
163 0210/1 2.00+ type_of_loader Boot loader identifier
164 0211/1 2.00+ loadflags Boot protocol option flags
165 0212/2 2.00+ setup_move_size Move to high memory size (used with hooks)
166 0214/4 2.00+ code32_start Boot loader hook (see below)
167 0218/4 2.00+ ramdisk_image initrd load address (set by boot loader)
168 021C/4 2.00+ ramdisk_size initrd size (set by boot loader)
169 0220/4 2.00+ bootsect_kludge DO NOT USE - for bootsect.S use only
170 0224/2 2.01+ heap_end_ptr Free memory after setup end
171 0226/2 N/A pad1 Unused
172 0228/4 2.02+ cmd_line_ptr 32-bit pointer to the kernel command line
173 022C/4 2.03+ ramdisk_max Highest legal initrd address
174 0230/4 2.05+ kernel_alignment Physical addr alignment required for kernel
175 0234/1 2.05+ relocatable_kernel Whether kernel is relocatable or not
176 0235/1 N/A pad2 Unused
177 0236/2 N/A pad3 Unused
178 0238/4 2.06+ cmdline_size Maximum size of the kernel command line
179 023C/4 2.07+ hardware_subarch Hardware subarchitecture
180 0240/8 2.07+ hardware_subarch_data Subarchitecture-specific data
181 0248/4 2.08+ payload_offset Offset of kernel payload
182 024C/4 2.08+ payload_length Length of kernel payload
183 0250/8 2.09+ setup_data 64-bit physical pointer to linked list
186 (1) For backwards compatibility, if the setup_sects field contains 0, the
189 (2) For boot protocol prior to 2.04, the upper two bytes of the syssize
190 field are unusable, which means the size of a bzImage kernel
191 cannot be determined.
193 If the "HdrS" (0x53726448) magic number is not found at offset 0x202,
194 the boot protocol version is "old". Loading an old kernel, the
195 following parameters should be assumed:
199 Real-mode kernel must be located at 0x90000.
201 Otherwise, the "version" field contains the protocol version,
202 e.g. protocol version 2.01 will contain 0x0201 in this field. When
203 setting fields in the header, you must make sure only to set fields
204 supported by the protocol version in use.
207 **** DETAILS OF HEADER FIELDS
209 For each field, some are information from the kernel to the bootloader
210 ("read"), some are expected to be filled out by the bootloader
211 ("write"), and some are expected to be read and modified by the
212 bootloader ("modify").
214 All general purpose boot loaders should write the fields marked
215 (obligatory). Boot loaders who want to load the kernel at a
216 nonstandard address should fill in the fields marked (reloc); other
217 boot loaders can ignore those fields.
219 The byte order of all fields is littleendian (this is x86, after all.)
221 Field name: setup_sects
226 The size of the setup code in 512-byte sectors. If this field is
227 0, the real value is 4. The real-mode code consists of the boot
228 sector (always one 512-byte sector) plus the setup code.
230 Field name: root_flags
231 Type: modify (optional)
235 If this field is nonzero, the root defaults to readonly. The use of
236 this field is deprecated; use the "ro" or "rw" options on the
237 command line instead.
241 Offset/size: 0x1f4/4 (protocol 2.04+) 0x1f4/2 (protocol ALL)
244 The size of the protected-mode code in units of 16-byte paragraphs.
245 For protocol versions older than 2.04 this field is only two bytes
246 wide, and therefore cannot be trusted for the size of a kernel if
247 the LOAD_HIGH flag is set.
250 Type: kernel internal
254 This field is obsolete.
257 Type: modify (obligatory)
260 Please see the section on SPECIAL COMMAND LINE OPTIONS.
263 Type: modify (optional)
267 The default root device device number. The use of this field is
268 deprecated, use the "root=" option on the command line instead.
270 Field name: boot_flag
275 Contains 0xAA55. This is the closest thing old Linux kernels have
283 Contains an x86 jump instruction, 0xEB followed by a signed offset
284 relative to byte 0x202. This can be used to determine the size of
292 Contains the magic number "HdrS" (0x53726448).
299 Contains the boot protocol version, in (major << 8)+minor format,
300 e.g. 0x0204 for version 2.04, and 0x0a11 for a hypothetical version
303 Field name: realmode_swtch
304 Type: modify (optional)
308 Boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
310 Field name: start_sys_seg
315 The load low segment (0x1000). Obsolete.
317 Field name: kernel_version
322 If set to a nonzero value, contains a pointer to a NUL-terminated
323 human-readable kernel version number string, less 0x200. This can
324 be used to display the kernel version to the user. This value
325 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:
332 0x1c00 < 15*0x200 (= 0x1e00) but
333 0x1c00 >= 14*0x200 (= 0x1c00)
335 0x1c00 >> 9 = 14, so the minimum value for setup_secs is 15.
337 Field name: type_of_loader
338 Type: write (obligatory)
342 If your boot loader has an assigned id (see table below), enter
343 0xTV here, where T is an identifier for the boot loader and V is
344 a version number. Otherwise, enter 0xFF here.
346 Assigned boot loader ids:
347 0 LILO (0x00 reserved for pre-2.00 bootloader)
349 2 bootsect-loader (0x20, all other values reserved)
359 Please contact <hpa@zytor.com> if you need a bootloader ID
362 Field name: loadflags
363 Type: modify (obligatory)
367 This field is a bitmask.
369 Bit 0 (read): LOADED_HIGH
370 - If 0, the protected-mode code is loaded at 0x10000.
371 - If 1, the protected-mode code is loaded at 0x100000.
373 Bit 5 (write): QUIET_FLAG
374 - If 0, print early messages.
375 - If 1, suppress early messages.
376 This requests to the kernel (decompressor and early
377 kernel) to not write early messages that require
378 accessing the display hardware directly.
380 Bit 6 (write): KEEP_SEGMENTS
382 - If 0, reload the segment registers in the 32bit entry point.
383 - If 1, do not reload the segment registers in the 32bit entry point.
384 Assume that %cs %ds %ss %es are all set to flat segments with
385 a base of 0 (or the equivalent for their environment).
387 Bit 7 (write): CAN_USE_HEAP
388 Set this bit to 1 to indicate that the value entered in the
389 heap_end_ptr is valid. If this field is clear, some setup code
390 functionality will be disabled.
392 Field name: setup_move_size
393 Type: modify (obligatory)
397 When using protocol 2.00 or 2.01, if the real mode kernel is not
398 loaded at 0x90000, it gets moved there later in the loading
399 sequence. Fill in this field if you want additional data (such as
400 the kernel command line) moved in addition to the real-mode kernel
403 The unit is bytes starting with the beginning of the boot sector.
405 This field is can be ignored when the protocol is 2.02 or higher, or
406 if the real-mode code is loaded at 0x90000.
408 Field name: code32_start
409 Type: modify (optional, reloc)
413 The address to jump to in protected mode. This defaults to the load
414 address of the kernel, and can be used by the boot loader to
415 determine the proper load address.
417 This field can be modified for two purposes:
419 1. as a boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
421 2. if a bootloader which does not install a hook loads a
422 relocatable kernel at a nonstandard address it will have to modify
423 this field to point to the load address.
425 Field name: ramdisk_image
426 Type: write (obligatory)
430 The 32-bit linear address of the initial ramdisk or ramfs. Leave at
431 zero if there is no initial ramdisk/ramfs.
433 Field name: ramdisk_size
434 Type: write (obligatory)
438 Size of the initial ramdisk or ramfs. Leave at zero if there is no
439 initial ramdisk/ramfs.
441 Field name: bootsect_kludge
442 Type: kernel internal
446 This field is obsolete.
448 Field name: heap_end_ptr
449 Type: write (obligatory)
453 Set this field to the offset (from the beginning of the real-mode
454 code) of the end of the setup stack/heap, minus 0x0200.
456 Field name: cmd_line_ptr
457 Type: write (obligatory)
461 Set this field to the linear address of the kernel command line.
462 The kernel command line can be located anywhere between the end of
463 the setup heap and 0xA0000; it does not have to be located in the
464 same 64K segment as the real-mode code itself.
466 Fill in this field even if your boot loader does not support a
467 command line, in which case you can point this to an empty string
468 (or better yet, to the string "auto".) If this field is left at
469 zero, the kernel will assume that your boot loader does not support
472 Field name: ramdisk_max
477 The maximum address that may be occupied by the initial
478 ramdisk/ramfs contents. For boot protocols 2.02 or earlier, this
479 field is not present, and the maximum address is 0x37FFFFFF. (This
480 address is defined as the address of the highest safe byte, so if
481 your ramdisk is exactly 131072 bytes long and this field is
482 0x37FFFFFF, you can start your ramdisk at 0x37FE0000.)
484 Field name: kernel_alignment
489 Alignment unit required by the kernel (if relocatable_kernel is true.)
491 Field name: relocatable_kernel
496 If this field is nonzero, the protected-mode part of the kernel can
497 be loaded at any address that satisfies the kernel_alignment field.
498 After loading, the boot loader must set the code32_start field to
499 point to the loaded code, or to a boot loader hook.
501 Field name: cmdline_size
506 The maximum size of the command line without the terminating
507 zero. This means that the command line can contain at most
508 cmdline_size characters. With protocol version 2.05 and earlier, the
509 maximum size was 255.
511 Field name: hardware_subarch
512 Type: write (optional, defaults to x86/PC)
516 In a paravirtualized environment the hardware low level architectural
517 pieces such as interrupt handling, page table handling, and
518 accessing process control registers needs to be done differently.
520 This field allows the bootloader to inform the kernel we are in one
521 one of those environments.
523 0x00000000 The default x86/PC environment
527 Field name: hardware_subarch_data
528 Type: write (subarch-dependent)
532 A pointer to data that is specific to hardware subarch
533 This field is currently unused for the default x86/PC environment,
536 Field name: payload_offset
541 If non-zero then this field contains the offset from the beginning
542 of the protected-mode code to the payload.
544 The payload may be compressed. The format of both the compressed and
545 uncompressed data should be determined using the standard magic
546 numbers. The currently supported compression formats are gzip
547 (magic numbers 1F 8B or 1F 9E), bzip2 (magic number 42 5A) and LZMA
548 (magic number 5D 00). The uncompressed payload is currently always ELF
549 (magic number 7F 45 4C 46).
551 Field name: payload_length
556 The length of the payload.
558 Field name: setup_data
559 Type: write (special)
563 The 64-bit physical pointer to NULL terminated single linked list of
564 struct setup_data. This is used to define a more extensible boot
565 parameters passing mechanism. The definition of struct setup_data is
575 Where, the next is a 64-bit physical pointer to the next node of
576 linked list, the next field of the last node is 0; the type is used
577 to identify the contents of data; the len is the length of data
578 field; the data holds the real payload.
580 This list may be modified at a number of points during the bootup
581 process. Therefore, when modifying this list one should always make
582 sure to consider the case where the linked list already contains
586 **** THE IMAGE CHECKSUM
588 From boot protocol version 2.08 onwards the CRC-32 is calculated over
589 the entire file using the characteristic polynomial 0x04C11DB7 and an
590 initial remainder of 0xffffffff. The checksum is appended to the
591 file; therefore the CRC of the file up to the limit specified in the
592 syssize field of the header is always 0.
595 **** THE KERNEL COMMAND LINE
597 The kernel command line has become an important way for the boot
598 loader to communicate with the kernel. Some of its options are also
599 relevant to the boot loader itself, see "special command line options"
602 The kernel command line is a null-terminated string. The maximum
603 length can be retrieved from the field cmdline_size. Before protocol
604 version 2.06, the maximum was 255 characters. A string that is too
605 long will be automatically truncated by the kernel.
607 If the boot protocol version is 2.02 or later, the address of the
608 kernel command line is given by the header field cmd_line_ptr (see
609 above.) This address can be anywhere between the end of the setup
612 If the protocol version is *not* 2.02 or higher, the kernel
613 command line is entered using the following protocol:
615 At offset 0x0020 (word), "cmd_line_magic", enter the magic
618 At offset 0x0022 (word), "cmd_line_offset", enter the offset
619 of the kernel command line (relative to the start of the
622 The kernel command line *must* be within the memory region
623 covered by setup_move_size, so you may need to adjust this
627 **** MEMORY LAYOUT OF THE REAL-MODE CODE
629 The real-mode code requires a stack/heap to be set up, as well as
630 memory allocated for the kernel command line. This needs to be done
631 in the real-mode accessible memory in bottom megabyte.
633 It should be noted that modern machines often have a sizable Extended
634 BIOS Data Area (EBDA). As a result, it is advisable to use as little
635 of the low megabyte as possible.
637 Unfortunately, under the following circumstances the 0x90000 memory
638 segment has to be used:
640 - When loading a zImage kernel ((loadflags & 0x01) == 0).
641 - When loading a 2.01 or earlier boot protocol kernel.
643 -> For the 2.00 and 2.01 boot protocols, the real-mode code
644 can be loaded at another address, but it is internally
645 relocated to 0x90000. For the "old" protocol, the
646 real-mode code must be loaded at 0x90000.
648 When loading at 0x90000, avoid using memory above 0x9a000.
650 For boot protocol 2.02 or higher, the command line does not have to be
651 located in the same 64K segment as the real-mode setup code; it is
652 thus permitted to give the stack/heap the full 64K segment and locate
653 the command line above it.
655 The kernel command line should not be located below the real-mode
656 code, nor should it be located in high memory.
659 **** SAMPLE BOOT CONFIGURATION
661 As a sample configuration, assume the following layout of the real
664 When loading below 0x90000, use the entire segment:
666 0x0000-0x7fff Real mode kernel
667 0x8000-0xdfff Stack and heap
668 0xe000-0xffff Kernel command line
670 When loading at 0x90000 OR the protocol version is 2.01 or earlier:
672 0x0000-0x7fff Real mode kernel
673 0x8000-0x97ff Stack and heap
674 0x9800-0x9fff Kernel command line
676 Such a boot loader should enter the following fields in the header:
678 unsigned long base_ptr; /* base address for real-mode segment */
680 if ( setup_sects == 0 ) {
684 if ( protocol >= 0x0200 ) {
685 type_of_loader = <type code>;
686 if ( loading_initrd ) {
687 ramdisk_image = <initrd_address>;
688 ramdisk_size = <initrd_size>;
691 if ( protocol >= 0x0202 && loadflags & 0x01 )
696 if ( protocol >= 0x0201 ) {
697 heap_end_ptr = heap_end - 0x200;
698 loadflags |= 0x80; /* CAN_USE_HEAP */
701 if ( protocol >= 0x0202 ) {
702 cmd_line_ptr = base_ptr + heap_end;
703 strcpy(cmd_line_ptr, cmdline);
705 cmd_line_magic = 0xA33F;
706 cmd_line_offset = heap_end;
707 setup_move_size = heap_end + strlen(cmdline)+1;
708 strcpy(base_ptr+cmd_line_offset, cmdline);
711 /* Very old kernel */
715 cmd_line_magic = 0xA33F;
716 cmd_line_offset = heap_end;
718 /* A very old kernel MUST have its real-mode code
721 if ( base_ptr != 0x90000 ) {
722 /* Copy the real-mode kernel */
723 memcpy(0x90000, base_ptr, (setup_sects+1)*512);
724 base_ptr = 0x90000; /* Relocated */
727 strcpy(0x90000+cmd_line_offset, cmdline);
729 /* It is recommended to clear memory up to the 32K mark */
730 memset(0x90000 + (setup_sects+1)*512, 0,
731 (64-(setup_sects+1))*512);
735 **** LOADING THE REST OF THE KERNEL
737 The 32-bit (non-real-mode) kernel starts at offset (setup_sects+1)*512
738 in the kernel file (again, if setup_sects == 0 the real value is 4.)
739 It should be loaded at address 0x10000 for Image/zImage kernels and
740 0x100000 for bzImage kernels.
742 The kernel is a bzImage kernel if the protocol >= 2.00 and the 0x01
743 bit (LOAD_HIGH) in the loadflags field is set:
745 is_bzImage = (protocol >= 0x0200) && (loadflags & 0x01);
746 load_address = is_bzImage ? 0x100000 : 0x10000;
748 Note that Image/zImage kernels can be up to 512K in size, and thus use
749 the entire 0x10000-0x90000 range of memory. This means it is pretty
750 much a requirement for these kernels to load the real-mode part at
751 0x90000. bzImage kernels allow much more flexibility.
754 **** SPECIAL COMMAND LINE OPTIONS
756 If the command line provided by the boot loader is entered by the
757 user, the user may expect the following command line options to work.
758 They should normally not be deleted from the kernel command line even
759 though not all of them are actually meaningful to the kernel. Boot
760 loader authors who need additional command line options for the boot
761 loader itself should get them registered in
762 Documentation/kernel-parameters.txt to make sure they will not
763 conflict with actual kernel options now or in the future.
766 <mode> here is either an integer (in C notation, either
767 decimal, octal, or hexadecimal) or one of the strings
768 "normal" (meaning 0xFFFF), "ext" (meaning 0xFFFE) or "ask"
769 (meaning 0xFFFD). This value should be entered into the
770 vid_mode field, as it is used by the kernel before the command
774 <size> is an integer in C notation optionally followed by
775 (case insensitive) K, M, G, T, P or E (meaning << 10, << 20,
776 << 30, << 40, << 50 or << 60). This specifies the end of
777 memory to the kernel. This affects the possible placement of
778 an initrd, since an initrd should be placed near end of
779 memory. Note that this is an option to *both* the kernel and
783 An initrd should be loaded. The meaning of <file> is
784 obviously bootloader-dependent, and some boot loaders
785 (e.g. LILO) do not have such a command.
787 In addition, some boot loaders add the following options to the
788 user-specified command line:
791 The boot image which was loaded. Again, the meaning of <file>
792 is obviously bootloader-dependent.
795 The kernel was booted without explicit user intervention.
797 If these options are added by the boot loader, it is highly
798 recommended that they are located *first*, before the user-specified
799 or configuration-specified command line. Otherwise, "init=/bin/sh"
800 gets confused by the "auto" option.
803 **** RUNNING THE KERNEL
805 The kernel is started by jumping to the kernel entry point, which is
806 located at *segment* offset 0x20 from the start of the real mode
807 kernel. This means that if you loaded your real-mode kernel code at
808 0x90000, the kernel entry point is 9020:0000.
810 At entry, ds = es = ss should point to the start of the real-mode
811 kernel code (0x9000 if the code is loaded at 0x90000), sp should be
812 set up properly, normally pointing to the top of the heap, and
813 interrupts should be disabled. Furthermore, to guard against bugs in
814 the kernel, it is recommended that the boot loader sets fs = gs = ds =
817 In our example from above, we would do:
819 /* Note: in the case of the "old" kernel protocol, base_ptr must
820 be == 0x90000 at this point; see the previous sample code */
824 cli(); /* Enter with interrupts disabled! */
826 /* Set up the real-mode kernel stack */
830 _DS = _ES = _FS = _GS = seg;
831 jmp_far(seg+0x20, 0); /* Run the kernel */
833 If your boot sector accesses a floppy drive, it is recommended to
834 switch off the floppy motor before running the kernel, since the
835 kernel boot leaves interrupts off and thus the motor will not be
836 switched off, especially if the loaded kernel has the floppy driver as
837 a demand-loaded module!
840 **** ADVANCED BOOT LOADER HOOKS
842 If the boot loader runs in a particularly hostile environment (such as
843 LOADLIN, which runs under DOS) it may be impossible to follow the
844 standard memory location requirements. Such a boot loader may use the
845 following hooks that, if set, are invoked by the kernel at the
846 appropriate time. The use of these hooks should probably be
847 considered an absolutely last resort!
849 IMPORTANT: All the hooks are required to preserve %esp, %ebp, %esi and
850 %edi across invocation.
853 A 16-bit real mode far subroutine invoked immediately before
854 entering protected mode. The default routine disables NMI, so
855 your routine should probably do so, too.
858 A 32-bit flat-mode routine *jumped* to immediately after the
859 transition to protected mode, but before the kernel is
860 uncompressed. No segments, except CS, are guaranteed to be
861 set up (current kernels do, but older ones do not); you should
862 set them up to BOOT_DS (0x18) yourself.
864 After completing your hook, you should jump to the address
865 that was in this field before your boot loader overwrote it
866 (relocated, if appropriate.)
869 **** 32-bit BOOT PROTOCOL
871 For machine with some new BIOS other than legacy BIOS, such as EFI,
872 LinuxBIOS, etc, and kexec, the 16-bit real mode setup code in kernel
873 based on legacy BIOS can not be used, so a 32-bit boot protocol needs
876 In 32-bit boot protocol, the first step in loading a Linux kernel
877 should be to setup the boot parameters (struct boot_params,
878 traditionally known as "zero page"). The memory for struct boot_params
879 should be allocated and initialized to all zero. Then the setup header
880 from offset 0x01f1 of kernel image on should be loaded into struct
881 boot_params and examined. The end of setup header can be calculated as
884 0x0202 + byte value at offset 0x0201
886 In addition to read/modify/write the setup header of the struct
887 boot_params as that of 16-bit boot protocol, the boot loader should
888 also fill the additional fields of the struct boot_params as that
889 described in zero-page.txt.
891 After setupping the struct boot_params, the boot loader can load the
892 32/64-bit kernel in the same way as that of 16-bit boot protocol.
894 In 32-bit boot protocol, the kernel is started by jumping to the
895 32-bit kernel entry point, which is the start address of loaded
898 At entry, the CPU must be in 32-bit protected mode with paging
899 disabled; a GDT must be loaded with the descriptors for selectors
900 __BOOT_CS(0x10) and __BOOT_DS(0x18); both descriptors must be 4G flat
901 segment; __BOOS_CS must have execute/read permission, and __BOOT_DS
902 must have read/write permission; CS must be __BOOT_CS and DS, ES, SS
903 must be __BOOT_DS; interrupt must be disabled; %esi must hold the base
904 address of the struct boot_params; %ebp, %edi and %ebx must be zero.