1 THE LINUX/I386 BOOT PROTOCOL
2 ----------------------------
4 H. Peter Anvin <hpa@zytor.com>
7 On the i386 platform, the Linux kernel uses a rather complicated boot
8 convention. This has evolved partially due to historical aspects, as
9 well as the desire in the early days to have the kernel itself be a
10 bootable image, the complicated PC memory model and due to changed
11 expectations in the PC industry caused by the effective demise of
12 real-mode DOS as a mainstream operating system.
14 Currently, the following versions of the Linux/i386 boot protocol exist.
16 Old kernels: zImage/Image support only. Some very early kernels
17 may not even support a command line.
19 Protocol 2.00: (Kernel 1.3.73) Added bzImage and initrd support, as
20 well as a formalized way to communicate between the
21 boot loader and the kernel. setup.S made relocatable,
22 although the traditional setup area still assumed
25 Protocol 2.01: (Kernel 1.3.76) Added a heap overrun warning.
27 Protocol 2.02: (Kernel 2.4.0-test3-pre3) New command line protocol.
28 Lower the conventional memory ceiling. No overwrite
29 of the traditional setup area, thus making booting
30 safe for systems which use the EBDA from SMM or 32-bit
31 BIOS entry points. zImage deprecated but still
34 Protocol 2.03: (Kernel 2.4.18-pre1) Explicitly makes the highest possible
35 initrd address available to the bootloader.
37 Protocol 2.04: (Kernel 2.6.14) Extend the syssize field to four bytes.
39 Protocol 2.05: (Kernel 2.6.20) Make protected mode kernel relocatable.
40 Introduce relocatable_kernel and kernel_alignment fields.
42 Protocol 2.06: (Kernel 2.6.22) Added a field that contains the size of
45 Protocol 2.09: (kernel 2.6.26) Added a field of 64-bit physical
46 pointer to single linked list of struct setup_data.
50 The traditional memory map for the kernel loader, used for Image or
51 zImage kernels, typically looks like:
54 0A0000 +------------------------+
55 | Reserved for BIOS | Do not use. Reserved for BIOS EBDA.
56 09A000 +------------------------+
58 | Stack/heap | For use by the kernel real-mode code.
59 098000 +------------------------+
60 | Kernel setup | The kernel real-mode code.
61 090200 +------------------------+
62 | Kernel boot sector | The kernel legacy boot sector.
63 090000 +------------------------+
64 | Protected-mode kernel | The bulk of the kernel image.
65 010000 +------------------------+
66 | Boot loader | <- Boot sector entry point 0000:7C00
67 001000 +------------------------+
68 | Reserved for MBR/BIOS |
69 000800 +------------------------+
70 | Typically used by MBR |
71 000600 +------------------------+
73 000000 +------------------------+
76 When using bzImage, the protected-mode kernel was relocated to
77 0x100000 ("high memory"), and the kernel real-mode block (boot sector,
78 setup, and stack/heap) was made relocatable to any address between
79 0x10000 and end of low memory. Unfortunately, in protocols 2.00 and
80 2.01 the 0x90000+ memory range is still used internally by the kernel;
81 the 2.02 protocol resolves that problem.
83 It is desirable to keep the "memory ceiling" -- the highest point in
84 low memory touched by the boot loader -- as low as possible, since
85 some newer BIOSes have begun to allocate some rather large amounts of
86 memory, called the Extended BIOS Data Area, near the top of low
87 memory. The boot loader should use the "INT 12h" BIOS call to verify
88 how much low memory is available.
90 Unfortunately, if INT 12h reports that the amount of memory is too
91 low, there is usually nothing the boot loader can do but to report an
92 error to the user. The boot loader should therefore be designed to
93 take up as little space in low memory as it reasonably can. For
94 zImage or old bzImage kernels, which need data written into the
95 0x90000 segment, the boot loader should make sure not to use memory
96 above the 0x9A000 point; too many BIOSes will break above that point.
98 For a modern bzImage kernel with boot protocol version >= 2.02, a
99 memory layout like the following is suggested:
102 | Protected-mode kernel |
103 100000 +------------------------+
105 0A0000 +------------------------+
106 | Reserved for BIOS | Leave as much as possible unused
108 | Command line | (Can also be below the X+10000 mark)
109 X+10000 +------------------------+
110 | Stack/heap | For use by the kernel real-mode code.
111 X+08000 +------------------------+
112 | Kernel setup | The kernel real-mode code.
113 | Kernel boot sector | The kernel legacy boot sector.
114 X +------------------------+
115 | Boot loader | <- Boot sector entry point 0000:7C00
116 001000 +------------------------+
117 | Reserved for MBR/BIOS |
118 000800 +------------------------+
119 | Typically used by MBR |
120 000600 +------------------------+
122 000000 +------------------------+
124 ... where the address X is as low as the design of the boot loader
128 **** THE REAL-MODE KERNEL HEADER
130 In the following text, and anywhere in the kernel boot sequence, "a
131 sector" refers to 512 bytes. It is independent of the actual sector
132 size of the underlying medium.
134 The first step in loading a Linux kernel should be to load the
135 real-mode code (boot sector and setup code) and then examine the
136 following header at offset 0x01f1. The real-mode code can total up to
137 32K, although the boot loader may choose to load only the first two
138 sectors (1K) and then examine the bootup sector size.
140 The header looks like:
142 Offset Proto Name Meaning
145 01F1/1 ALL(1 setup_sects The size of the setup in sectors
146 01F2/2 ALL root_flags If set, the root is mounted readonly
147 01F4/4 2.04+(2 syssize The size of the 32-bit code in 16-byte paras
148 01F8/2 ALL ram_size DO NOT USE - for bootsect.S use only
149 01FA/2 ALL vid_mode Video mode control
150 01FC/2 ALL root_dev Default root device number
151 01FE/2 ALL boot_flag 0xAA55 magic number
152 0200/2 2.00+ jump Jump instruction
153 0202/4 2.00+ header Magic signature "HdrS"
154 0206/2 2.00+ version Boot protocol version supported
155 0208/4 2.00+ realmode_swtch Boot loader hook (see below)
156 020C/2 2.00+ start_sys The load-low segment (0x1000) (obsolete)
157 020E/2 2.00+ kernel_version Pointer to kernel version string
158 0210/1 2.00+ type_of_loader Boot loader identifier
159 0211/1 2.00+ loadflags Boot protocol option flags
160 0212/2 2.00+ setup_move_size Move to high memory size (used with hooks)
161 0214/4 2.00+ code32_start Boot loader hook (see below)
162 0218/4 2.00+ ramdisk_image initrd load address (set by boot loader)
163 021C/4 2.00+ ramdisk_size initrd size (set by boot loader)
164 0220/4 2.00+ bootsect_kludge DO NOT USE - for bootsect.S use only
165 0224/2 2.01+ heap_end_ptr Free memory after setup end
166 0226/2 N/A pad1 Unused
167 0228/4 2.02+ cmd_line_ptr 32-bit pointer to the kernel command line
168 022C/4 2.03+ initrd_addr_max Highest legal initrd address
169 0230/4 2.05+ kernel_alignment Physical addr alignment required for kernel
170 0234/1 2.05+ relocatable_kernel Whether kernel is relocatable or not
171 0235/3 N/A pad2 Unused
172 0238/4 2.06+ cmdline_size Maximum size of the kernel command line
173 023C/4 2.07+ hardware_subarch Hardware subarchitecture
174 0240/8 2.07+ hardware_subarch_data Subarchitecture-specific data
175 0248/4 2.08+ payload_offset Offset of kernel payload
176 024C/4 2.08+ payload_length Length of kernel payload
177 0250/8 2.09+ setup_data 64-bit physical pointer to linked list
180 (1) For backwards compatibility, if the setup_sects field contains 0, the
183 (2) For boot protocol prior to 2.04, the upper two bytes of the syssize
184 field are unusable, which means the size of a bzImage kernel
185 cannot be determined.
187 If the "HdrS" (0x53726448) magic number is not found at offset 0x202,
188 the boot protocol version is "old". Loading an old kernel, the
189 following parameters should be assumed:
193 Real-mode kernel must be located at 0x90000.
195 Otherwise, the "version" field contains the protocol version,
196 e.g. protocol version 2.01 will contain 0x0201 in this field. When
197 setting fields in the header, you must make sure only to set fields
198 supported by the protocol version in use.
201 **** DETAILS OF HEADER FIELDS
203 For each field, some are information from the kernel to the bootloader
204 ("read"), some are expected to be filled out by the bootloader
205 ("write"), and some are expected to be read and modified by the
206 bootloader ("modify").
208 All general purpose boot loaders should write the fields marked
209 (obligatory). Boot loaders who want to load the kernel at a
210 nonstandard address should fill in the fields marked (reloc); other
211 boot loaders can ignore those fields.
213 The byte order of all fields is littleendian (this is x86, after all.)
215 Field name: setup_sects
220 The size of the setup code in 512-byte sectors. If this field is
221 0, the real value is 4. The real-mode code consists of the boot
222 sector (always one 512-byte sector) plus the setup code.
224 Field name: root_flags
225 Type: modify (optional)
229 If this field is nonzero, the root defaults to readonly. The use of
230 this field is deprecated; use the "ro" or "rw" options on the
231 command line instead.
235 Offset/size: 0x1f4/4 (protocol 2.04+) 0x1f4/2 (protocol ALL)
238 The size of the protected-mode code in units of 16-byte paragraphs.
239 For protocol versions older than 2.04 this field is only two bytes
240 wide, and therefore cannot be trusted for the size of a kernel if
241 the LOAD_HIGH flag is set.
244 Type: kernel internal
248 This field is obsolete.
251 Type: modify (obligatory)
254 Please see the section on SPECIAL COMMAND LINE OPTIONS.
257 Type: modify (optional)
261 The default root device device number. The use of this field is
262 deprecated, use the "root=" option on the command line instead.
264 Field name: boot_flag
269 Contains 0xAA55. This is the closest thing old Linux kernels have
277 Contains an x86 jump instruction, 0xEB followed by a signed offset
278 relative to byte 0x202. This can be used to determine the size of
286 Contains the magic number "HdrS" (0x53726448).
293 Contains the boot protocol version, in (major << 8)+minor format,
294 e.g. 0x0204 for version 2.04, and 0x0a11 for a hypothetical version
297 Field name: readmode_swtch
298 Type: modify (optional)
302 Boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
304 Field name: start_sys
309 The load low segment (0x1000). Obsolete.
311 Field name: kernel_version
316 If set to a nonzero value, contains a pointer to a NUL-terminated
317 human-readable kernel version number string, less 0x200. This can
318 be used to display the kernel version to the user. This value
319 should be less than (0x200*setup_sects).
321 For example, if this value is set to 0x1c00, the kernel version
322 number string can be found at offset 0x1e00 in the kernel file.
323 This is a valid value if and only if the "setup_sects" field
324 contains the value 15 or higher, as:
326 0x1c00 < 15*0x200 (= 0x1e00) but
327 0x1c00 >= 14*0x200 (= 0x1c00)
329 0x1c00 >> 9 = 14, so the minimum value for setup_secs is 15.
331 Field name: type_of_loader
332 Type: write (obligatory)
336 If your boot loader has an assigned id (see table below), enter
337 0xTV here, where T is an identifier for the boot loader and V is
338 a version number. Otherwise, enter 0xFF here.
340 Assigned boot loader ids:
341 0 LILO (0x00 reserved for pre-2.00 bootloader)
343 2 bootsect-loader (0x20, all other values reserved)
353 Please contact <hpa@zytor.com> if you need a bootloader ID
356 Field name: loadflags
357 Type: modify (obligatory)
361 This field is a bitmask.
363 Bit 0 (read): LOADED_HIGH
364 - If 0, the protected-mode code is loaded at 0x10000.
365 - If 1, the protected-mode code is loaded at 0x100000.
367 Bit 6 (write): KEEP_SEGMENTS
369 - if 0, reload the segment registers in the 32bit entry point.
370 - if 1, do not reload the segment registers in the 32bit entry point.
371 Assume that %cs %ds %ss %es are all set to flat segments with
372 a base of 0 (or the equivalent for their environment).
374 Bit 7 (write): CAN_USE_HEAP
375 Set this bit to 1 to indicate that the value entered in the
376 heap_end_ptr is valid. If this field is clear, some setup code
377 functionality will be disabled.
379 Field name: setup_move_size
380 Type: modify (obligatory)
384 When using protocol 2.00 or 2.01, if the real mode kernel is not
385 loaded at 0x90000, it gets moved there later in the loading
386 sequence. Fill in this field if you want additional data (such as
387 the kernel command line) moved in addition to the real-mode kernel
390 The unit is bytes starting with the beginning of the boot sector.
392 This field is can be ignored when the protocol is 2.02 or higher, or
393 if the real-mode code is loaded at 0x90000.
395 Field name: code32_start
396 Type: modify (optional, reloc)
400 The address to jump to in protected mode. This defaults to the load
401 address of the kernel, and can be used by the boot loader to
402 determine the proper load address.
404 This field can be modified for two purposes:
406 1. as a boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
408 2. if a bootloader which does not install a hook loads a
409 relocatable kernel at a nonstandard address it will have to modify
410 this field to point to the load address.
412 Field name: ramdisk_image
413 Type: write (obligatory)
417 The 32-bit linear address of the initial ramdisk or ramfs. Leave at
418 zero if there is no initial ramdisk/ramfs.
420 Field name: ramdisk_size
421 Type: write (obligatory)
425 Size of the initial ramdisk or ramfs. Leave at zero if there is no
426 initial ramdisk/ramfs.
428 Field name: bootsect_kludge
429 Type: kernel internal
433 This field is obsolete.
435 Field name: heap_end_ptr
436 Type: write (obligatory)
440 Set this field to the offset (from the beginning of the real-mode
441 code) of the end of the setup stack/heap, minus 0x0200.
443 Field name: cmd_line_ptr
444 Type: write (obligatory)
448 Set this field to the linear address of the kernel command line.
449 The kernel command line can be located anywhere between the end of
450 the setup heap and 0xA0000; it does not have to be located in the
451 same 64K segment as the real-mode code itself.
453 Fill in this field even if your boot loader does not support a
454 command line, in which case you can point this to an empty string
455 (or better yet, to the string "auto".) If this field is left at
456 zero, the kernel will assume that your boot loader does not support
459 Field name: initrd_addr_max
464 The maximum address that may be occupied by the initial
465 ramdisk/ramfs contents. For boot protocols 2.02 or earlier, this
466 field is not present, and the maximum address is 0x37FFFFFF. (This
467 address is defined as the address of the highest safe byte, so if
468 your ramdisk is exactly 131072 bytes long and this field is
469 0x37FFFFFF, you can start your ramdisk at 0x37FE0000.)
471 Field name: kernel_alignment
476 Alignment unit required by the kernel (if relocatable_kernel is true.)
478 Field name: relocatable_kernel
483 If this field is nonzero, the protected-mode part of the kernel can
484 be loaded at any address that satisfies the kernel_alignment field.
485 After loading, the boot loader must set the code32_start field to
486 point to the loaded code, or to a boot loader hook.
488 Field name: cmdline_size
493 The maximum size of the command line without the terminating
494 zero. This means that the command line can contain at most
495 cmdline_size characters. With protocol version 2.05 and earlier, the
496 maximum size was 255.
498 Field name: hardware_subarch
503 In a paravirtualized environment the hardware low level architectural
504 pieces such as interrupt handling, page table handling, and
505 accessing process control registers needs to be done differently.
507 This field allows the bootloader to inform the kernel we are in one
508 one of those environments.
510 0x00000000 The default x86/PC environment
514 Field name: hardware_subarch_data
519 A pointer to data that is specific to hardware subarch
521 Field name: payload_offset
526 If non-zero then this field contains the offset from the end of the
527 real-mode code to the payload.
529 The payload may be compressed. The format of both the compressed and
530 uncompressed data should be determined using the standard magic
531 numbers. Currently only gzip compressed ELF is used.
533 Field name: payload_length
538 The length of the payload.
540 **** THE IMAGE CHECKSUM
542 From boot protocol version 2.08 onwards the CRC-32 is calculated over
543 the entire file using the characteristic polynomial 0x04C11DB7 and an
544 initial remainder of 0xffffffff. The checksum is appended to the
545 file; therefore the CRC of the file up to the limit specified in the
546 syssize field of the header is always 0.
548 **** THE KERNEL COMMAND LINE
550 The kernel command line has become an important way for the boot
551 loader to communicate with the kernel. Some of its options are also
552 relevant to the boot loader itself, see "special command line options"
555 The kernel command line is a null-terminated string. The maximum
556 length can be retrieved from the field cmdline_size. Before protocol
557 version 2.06, the maximum was 255 characters. A string that is too
558 long will be automatically truncated by the kernel.
560 If the boot protocol version is 2.02 or later, the address of the
561 kernel command line is given by the header field cmd_line_ptr (see
562 above.) This address can be anywhere between the end of the setup
565 If the protocol version is *not* 2.02 or higher, the kernel
566 command line is entered using the following protocol:
568 At offset 0x0020 (word), "cmd_line_magic", enter the magic
571 At offset 0x0022 (word), "cmd_line_offset", enter the offset
572 of the kernel command line (relative to the start of the
575 The kernel command line *must* be within the memory region
576 covered by setup_move_size, so you may need to adjust this
579 Field name: setup_data
580 Type: write (obligatory)
584 The 64-bit physical pointer to NULL terminated single linked list of
585 struct setup_data. This is used to define a more extensible boot
586 parameters passing mechanism. The definition of struct setup_data is
596 Where, the next is a 64-bit physical pointer to the next node of
597 linked list, the next field of the last node is 0; the type is used
598 to identify the contents of data; the len is the length of data
599 field; the data holds the real payload.
602 **** MEMORY LAYOUT OF THE REAL-MODE CODE
604 The real-mode code requires a stack/heap to be set up, as well as
605 memory allocated for the kernel command line. This needs to be done
606 in the real-mode accessible memory in bottom megabyte.
608 It should be noted that modern machines often have a sizable Extended
609 BIOS Data Area (EBDA). As a result, it is advisable to use as little
610 of the low megabyte as possible.
612 Unfortunately, under the following circumstances the 0x90000 memory
613 segment has to be used:
615 - When loading a zImage kernel ((loadflags & 0x01) == 0).
616 - When loading a 2.01 or earlier boot protocol kernel.
618 -> For the 2.00 and 2.01 boot protocols, the real-mode code
619 can be loaded at another address, but it is internally
620 relocated to 0x90000. For the "old" protocol, the
621 real-mode code must be loaded at 0x90000.
623 When loading at 0x90000, avoid using memory above 0x9a000.
625 For boot protocol 2.02 or higher, the command line does not have to be
626 located in the same 64K segment as the real-mode setup code; it is
627 thus permitted to give the stack/heap the full 64K segment and locate
628 the command line above it.
630 The kernel command line should not be located below the real-mode
631 code, nor should it be located in high memory.
634 **** SAMPLE BOOT CONFIGURATION
636 As a sample configuration, assume the following layout of the real
639 When loading below 0x90000, use the entire segment:
641 0x0000-0x7fff Real mode kernel
642 0x8000-0xdfff Stack and heap
643 0xe000-0xffff Kernel command line
645 When loading at 0x90000 OR the protocol version is 2.01 or earlier:
647 0x0000-0x7fff Real mode kernel
648 0x8000-0x97ff Stack and heap
649 0x9800-0x9fff Kernel command line
651 Such a boot loader should enter the following fields in the header:
653 unsigned long base_ptr; /* base address for real-mode segment */
655 if ( setup_sects == 0 ) {
659 if ( protocol >= 0x0200 ) {
660 type_of_loader = <type code>;
661 if ( loading_initrd ) {
662 ramdisk_image = <initrd_address>;
663 ramdisk_size = <initrd_size>;
666 if ( protocol >= 0x0202 && loadflags & 0x01 )
671 if ( protocol >= 0x0201 ) {
672 heap_end_ptr = heap_end - 0x200;
673 loadflags |= 0x80; /* CAN_USE_HEAP */
676 if ( protocol >= 0x0202 ) {
677 cmd_line_ptr = base_ptr + heap_end;
678 strcpy(cmd_line_ptr, cmdline);
680 cmd_line_magic = 0xA33F;
681 cmd_line_offset = heap_end;
682 setup_move_size = heap_end + strlen(cmdline)+1;
683 strcpy(base_ptr+cmd_line_offset, cmdline);
686 /* Very old kernel */
690 cmd_line_magic = 0xA33F;
691 cmd_line_offset = heap_end;
693 /* A very old kernel MUST have its real-mode code
696 if ( base_ptr != 0x90000 ) {
697 /* Copy the real-mode kernel */
698 memcpy(0x90000, base_ptr, (setup_sects+1)*512);
699 base_ptr = 0x90000; /* Relocated */
702 strcpy(0x90000+cmd_line_offset, cmdline);
704 /* It is recommended to clear memory up to the 32K mark */
705 memset(0x90000 + (setup_sects+1)*512, 0,
706 (64-(setup_sects+1))*512);
710 **** LOADING THE REST OF THE KERNEL
712 The 32-bit (non-real-mode) kernel starts at offset (setup_sects+1)*512
713 in the kernel file (again, if setup_sects == 0 the real value is 4.)
714 It should be loaded at address 0x10000 for Image/zImage kernels and
715 0x100000 for bzImage kernels.
717 The kernel is a bzImage kernel if the protocol >= 2.00 and the 0x01
718 bit (LOAD_HIGH) in the loadflags field is set:
720 is_bzImage = (protocol >= 0x0200) && (loadflags & 0x01);
721 load_address = is_bzImage ? 0x100000 : 0x10000;
723 Note that Image/zImage kernels can be up to 512K in size, and thus use
724 the entire 0x10000-0x90000 range of memory. This means it is pretty
725 much a requirement for these kernels to load the real-mode part at
726 0x90000. bzImage kernels allow much more flexibility.
729 **** SPECIAL COMMAND LINE OPTIONS
731 If the command line provided by the boot loader is entered by the
732 user, the user may expect the following command line options to work.
733 They should normally not be deleted from the kernel command line even
734 though not all of them are actually meaningful to the kernel. Boot
735 loader authors who need additional command line options for the boot
736 loader itself should get them registered in
737 Documentation/kernel-parameters.txt to make sure they will not
738 conflict with actual kernel options now or in the future.
741 <mode> here is either an integer (in C notation, either
742 decimal, octal, or hexadecimal) or one of the strings
743 "normal" (meaning 0xFFFF), "ext" (meaning 0xFFFE) or "ask"
744 (meaning 0xFFFD). This value should be entered into the
745 vid_mode field, as it is used by the kernel before the command
749 <size> is an integer in C notation optionally followed by
750 (case insensitive) K, M, G, T, P or E (meaning << 10, << 20,
751 << 30, << 40, << 50 or << 60). This specifies the end of
752 memory to the kernel. This affects the possible placement of
753 an initrd, since an initrd should be placed near end of
754 memory. Note that this is an option to *both* the kernel and
758 An initrd should be loaded. The meaning of <file> is
759 obviously bootloader-dependent, and some boot loaders
760 (e.g. LILO) do not have such a command.
762 In addition, some boot loaders add the following options to the
763 user-specified command line:
766 The boot image which was loaded. Again, the meaning of <file>
767 is obviously bootloader-dependent.
770 The kernel was booted without explicit user intervention.
772 If these options are added by the boot loader, it is highly
773 recommended that they are located *first*, before the user-specified
774 or configuration-specified command line. Otherwise, "init=/bin/sh"
775 gets confused by the "auto" option.
778 **** RUNNING THE KERNEL
780 The kernel is started by jumping to the kernel entry point, which is
781 located at *segment* offset 0x20 from the start of the real mode
782 kernel. This means that if you loaded your real-mode kernel code at
783 0x90000, the kernel entry point is 9020:0000.
785 At entry, ds = es = ss should point to the start of the real-mode
786 kernel code (0x9000 if the code is loaded at 0x90000), sp should be
787 set up properly, normally pointing to the top of the heap, and
788 interrupts should be disabled. Furthermore, to guard against bugs in
789 the kernel, it is recommended that the boot loader sets fs = gs = ds =
792 In our example from above, we would do:
794 /* Note: in the case of the "old" kernel protocol, base_ptr must
795 be == 0x90000 at this point; see the previous sample code */
799 cli(); /* Enter with interrupts disabled! */
801 /* Set up the real-mode kernel stack */
805 _DS = _ES = _FS = _GS = seg;
806 jmp_far(seg+0x20, 0); /* Run the kernel */
808 If your boot sector accesses a floppy drive, it is recommended to
809 switch off the floppy motor before running the kernel, since the
810 kernel boot leaves interrupts off and thus the motor will not be
811 switched off, especially if the loaded kernel has the floppy driver as
812 a demand-loaded module!
815 **** ADVANCED BOOT LOADER HOOKS
817 If the boot loader runs in a particularly hostile environment (such as
818 LOADLIN, which runs under DOS) it may be impossible to follow the
819 standard memory location requirements. Such a boot loader may use the
820 following hooks that, if set, are invoked by the kernel at the
821 appropriate time. The use of these hooks should probably be
822 considered an absolutely last resort!
824 IMPORTANT: All the hooks are required to preserve %esp, %ebp, %esi and
825 %edi across invocation.
828 A 16-bit real mode far subroutine invoked immediately before
829 entering protected mode. The default routine disables NMI, so
830 your routine should probably do so, too.
833 A 32-bit flat-mode routine *jumped* to immediately after the
834 transition to protected mode, but before the kernel is
835 uncompressed. No segments, except CS, are guaranteed to be
836 set up (current kernels do, but older ones do not); you should
837 set them up to BOOT_DS (0x18) yourself.
839 After completing your hook, you should jump to the address
840 that was in this field before your boot loader overwrote it
841 (relocated, if appropriate.)
844 **** 32-bit BOOT PROTOCOL
846 For machine with some new BIOS other than legacy BIOS, such as EFI,
847 LinuxBIOS, etc, and kexec, the 16-bit real mode setup code in kernel
848 based on legacy BIOS can not be used, so a 32-bit boot protocol needs
851 In 32-bit boot protocol, the first step in loading a Linux kernel
852 should be to setup the boot parameters (struct boot_params,
853 traditionally known as "zero page"). The memory for struct boot_params
854 should be allocated and initialized to all zero. Then the setup header
855 from offset 0x01f1 of kernel image on should be loaded into struct
856 boot_params and examined. The end of setup header can be calculated as
859 0x0202 + byte value at offset 0x0201
861 In addition to read/modify/write the setup header of the struct
862 boot_params as that of 16-bit boot protocol, the boot loader should
863 also fill the additional fields of the struct boot_params as that
864 described in zero-page.txt.
866 After setupping the struct boot_params, the boot loader can load the
867 32/64-bit kernel in the same way as that of 16-bit boot protocol.
869 In 32-bit boot protocol, the kernel is started by jumping to the
870 32-bit kernel entry point, which is the start address of loaded
873 At entry, the CPU must be in 32-bit protected mode with paging
874 disabled; a GDT must be loaded with the descriptors for selectors
875 __BOOT_CS(0x10) and __BOOT_DS(0x18); both descriptors must be 4G flat
876 segment; __BOOS_CS must have execute/read permission, and __BOOT_DS
877 must have read/write permission; CS must be __BOOT_CS and DS, ES, SS
878 must be __BOOT_DS; interrupt must be disabled; %esi must hold the base
879 address of the struct boot_params; %ebp, %edi and %ebx must be zero.