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
43 the boot command line.
45 Protocol 2.07: (Kernel 2.6.24) Added paravirtualised boot protocol.
46 Introduced hardware_subarch and hardware_subarch_data
47 and KEEP_SEGMENTS flag in load_flags.
49 Protocol 2.08: (Kernel 2.6.26) Added crc32 checksum and ELF format
50 payload. Introduced payload_offset and payload length
51 fields to aid in locating the payload.
53 Protocol 2.09: (Kernel 2.6.26) Added a field of 64-bit physical
54 pointer to single linked list of struct setup_data.
58 The traditional memory map for the kernel loader, used for Image or
59 zImage kernels, typically looks like:
62 0A0000 +------------------------+
63 | Reserved for BIOS | Do not use. Reserved for BIOS EBDA.
64 09A000 +------------------------+
66 | Stack/heap | For use by the kernel real-mode code.
67 098000 +------------------------+
68 | Kernel setup | The kernel real-mode code.
69 090200 +------------------------+
70 | Kernel boot sector | The kernel legacy boot sector.
71 090000 +------------------------+
72 | Protected-mode kernel | The bulk of the kernel image.
73 010000 +------------------------+
74 | Boot loader | <- Boot sector entry point 0000:7C00
75 001000 +------------------------+
76 | Reserved for MBR/BIOS |
77 000800 +------------------------+
78 | Typically used by MBR |
79 000600 +------------------------+
81 000000 +------------------------+
84 When using bzImage, the protected-mode kernel was relocated to
85 0x100000 ("high memory"), and the kernel real-mode block (boot sector,
86 setup, and stack/heap) was made relocatable to any address between
87 0x10000 and end of low memory. Unfortunately, in protocols 2.00 and
88 2.01 the 0x90000+ memory range is still used internally by the kernel;
89 the 2.02 protocol resolves that problem.
91 It is desirable to keep the "memory ceiling" -- the highest point in
92 low memory touched by the boot loader -- as low as possible, since
93 some newer BIOSes have begun to allocate some rather large amounts of
94 memory, called the Extended BIOS Data Area, near the top of low
95 memory. The boot loader should use the "INT 12h" BIOS call to verify
96 how much low memory is available.
98 Unfortunately, if INT 12h reports that the amount of memory is too
99 low, there is usually nothing the boot loader can do but to report an
100 error to the user. The boot loader should therefore be designed to
101 take up as little space in low memory as it reasonably can. For
102 zImage or old bzImage kernels, which need data written into the
103 0x90000 segment, the boot loader should make sure not to use memory
104 above the 0x9A000 point; too many BIOSes will break above that point.
106 For a modern bzImage kernel with boot protocol version >= 2.02, a
107 memory layout like the following is suggested:
110 | Protected-mode kernel |
111 100000 +------------------------+
113 0A0000 +------------------------+
114 | Reserved for BIOS | Leave as much as possible unused
116 | Command line | (Can also be below the X+10000 mark)
117 X+10000 +------------------------+
118 | Stack/heap | For use by the kernel real-mode code.
119 X+08000 +------------------------+
120 | Kernel setup | The kernel real-mode code.
121 | Kernel boot sector | The kernel legacy boot sector.
122 X +------------------------+
123 | Boot loader | <- Boot sector entry point 0000:7C00
124 001000 +------------------------+
125 | Reserved for MBR/BIOS |
126 000800 +------------------------+
127 | Typically used by MBR |
128 000600 +------------------------+
130 000000 +------------------------+
132 ... where the address X is as low as the design of the boot loader
136 **** THE REAL-MODE KERNEL HEADER
138 In the following text, and anywhere in the kernel boot sequence, "a
139 sector" refers to 512 bytes. It is independent of the actual sector
140 size of the underlying medium.
142 The first step in loading a Linux kernel should be to load the
143 real-mode code (boot sector and setup code) and then examine the
144 following header at offset 0x01f1. The real-mode code can total up to
145 32K, although the boot loader may choose to load only the first two
146 sectors (1K) and then examine the bootup sector size.
148 The header looks like:
150 Offset Proto Name Meaning
153 01F1/1 ALL(1 setup_sects The size of the setup in sectors
154 01F2/2 ALL root_flags If set, the root is mounted readonly
155 01F4/4 2.04+(2 syssize The size of the 32-bit code in 16-byte paras
156 01F8/2 ALL ram_size DO NOT USE - for bootsect.S use only
157 01FA/2 ALL vid_mode Video mode control
158 01FC/2 ALL root_dev Default root device number
159 01FE/2 ALL boot_flag 0xAA55 magic number
160 0200/2 2.00+ jump Jump instruction
161 0202/4 2.00+ header Magic signature "HdrS"
162 0206/2 2.00+ version Boot protocol version supported
163 0208/4 2.00+ realmode_swtch Boot loader hook (see below)
164 020C/2 2.00+ start_sys The load-low segment (0x1000) (obsolete)
165 020E/2 2.00+ kernel_version Pointer to kernel version string
166 0210/1 2.00+ type_of_loader Boot loader identifier
167 0211/1 2.00+ loadflags Boot protocol option flags
168 0212/2 2.00+ setup_move_size Move to high memory size (used with hooks)
169 0214/4 2.00+ code32_start Boot loader hook (see below)
170 0218/4 2.00+ ramdisk_image initrd load address (set by boot loader)
171 021C/4 2.00+ ramdisk_size initrd size (set by boot loader)
172 0220/4 2.00+ bootsect_kludge DO NOT USE - for bootsect.S use only
173 0224/2 2.01+ heap_end_ptr Free memory after setup end
174 0226/2 N/A pad1 Unused
175 0228/4 2.02+ cmd_line_ptr 32-bit pointer to the kernel command line
176 022C/4 2.03+ initrd_addr_max Highest legal initrd address
177 0230/4 2.05+ kernel_alignment Physical addr alignment required for kernel
178 0234/1 2.05+ relocatable_kernel Whether kernel is relocatable or not
179 0235/3 N/A pad2 Unused
180 0238/4 2.06+ cmdline_size Maximum size of the kernel command line
181 023C/4 2.07+ hardware_subarch Hardware subarchitecture
182 0240/8 2.07+ hardware_subarch_data Subarchitecture-specific data
183 0248/4 2.08+ payload_offset Offset of kernel payload
184 024C/4 2.08+ payload_length Length of kernel payload
185 0250/8 2.09+ setup_data 64-bit physical pointer to linked list
188 (1) For backwards compatibility, if the setup_sects field contains 0, the
191 (2) For boot protocol prior to 2.04, the upper two bytes of the syssize
192 field are unusable, which means the size of a bzImage kernel
193 cannot be determined.
195 If the "HdrS" (0x53726448) magic number is not found at offset 0x202,
196 the boot protocol version is "old". Loading an old kernel, the
197 following parameters should be assumed:
201 Real-mode kernel must be located at 0x90000.
203 Otherwise, the "version" field contains the protocol version,
204 e.g. protocol version 2.01 will contain 0x0201 in this field. When
205 setting fields in the header, you must make sure only to set fields
206 supported by the protocol version in use.
209 **** DETAILS OF HEADER FIELDS
211 For each field, some are information from the kernel to the bootloader
212 ("read"), some are expected to be filled out by the bootloader
213 ("write"), and some are expected to be read and modified by the
214 bootloader ("modify").
216 All general purpose boot loaders should write the fields marked
217 (obligatory). Boot loaders who want to load the kernel at a
218 nonstandard address should fill in the fields marked (reloc); other
219 boot loaders can ignore those fields.
221 The byte order of all fields is littleendian (this is x86, after all.)
223 Field name: setup_sects
228 The size of the setup code in 512-byte sectors. If this field is
229 0, the real value is 4. The real-mode code consists of the boot
230 sector (always one 512-byte sector) plus the setup code.
232 Field name: root_flags
233 Type: modify (optional)
237 If this field is nonzero, the root defaults to readonly. The use of
238 this field is deprecated; use the "ro" or "rw" options on the
239 command line instead.
243 Offset/size: 0x1f4/4 (protocol 2.04+) 0x1f4/2 (protocol ALL)
246 The size of the protected-mode code in units of 16-byte paragraphs.
247 For protocol versions older than 2.04 this field is only two bytes
248 wide, and therefore cannot be trusted for the size of a kernel if
249 the LOAD_HIGH flag is set.
252 Type: kernel internal
256 This field is obsolete.
259 Type: modify (obligatory)
262 Please see the section on SPECIAL COMMAND LINE OPTIONS.
265 Type: modify (optional)
269 The default root device device number. The use of this field is
270 deprecated, use the "root=" option on the command line instead.
272 Field name: boot_flag
277 Contains 0xAA55. This is the closest thing old Linux kernels have
285 Contains an x86 jump instruction, 0xEB followed by a signed offset
286 relative to byte 0x202. This can be used to determine the size of
294 Contains the magic number "HdrS" (0x53726448).
301 Contains the boot protocol version, in (major << 8)+minor format,
302 e.g. 0x0204 for version 2.04, and 0x0a11 for a hypothetical version
305 Field name: readmode_swtch
306 Type: modify (optional)
310 Boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
312 Field name: start_sys
317 The load low segment (0x1000). Obsolete.
319 Field name: kernel_version
324 If set to a nonzero value, contains a pointer to a NUL-terminated
325 human-readable kernel version number string, less 0x200. This can
326 be used to display the kernel version to the user. This value
327 should be less than (0x200*setup_sects).
329 For example, if this value is set to 0x1c00, the kernel version
330 number string can be found at offset 0x1e00 in the kernel file.
331 This is a valid value if and only if the "setup_sects" field
332 contains the value 15 or higher, as:
334 0x1c00 < 15*0x200 (= 0x1e00) but
335 0x1c00 >= 14*0x200 (= 0x1c00)
337 0x1c00 >> 9 = 14, so the minimum value for setup_secs is 15.
339 Field name: type_of_loader
340 Type: write (obligatory)
344 If your boot loader has an assigned id (see table below), enter
345 0xTV here, where T is an identifier for the boot loader and V is
346 a version number. Otherwise, enter 0xFF here.
348 Assigned boot loader ids:
349 0 LILO (0x00 reserved for pre-2.00 bootloader)
351 2 bootsect-loader (0x20, all other values reserved)
361 Please contact <hpa@zytor.com> if you need a bootloader ID
364 Field name: loadflags
365 Type: modify (obligatory)
369 This field is a bitmask.
371 Bit 0 (read): LOADED_HIGH
372 - If 0, the protected-mode code is loaded at 0x10000.
373 - If 1, the protected-mode code is loaded at 0x100000.
375 Bit 6 (write): KEEP_SEGMENTS
377 - if 0, reload the segment registers in the 32bit entry point.
378 - if 1, do not reload the segment registers in the 32bit entry point.
379 Assume that %cs %ds %ss %es are all set to flat segments with
380 a base of 0 (or the equivalent for their environment).
382 Bit 7 (write): CAN_USE_HEAP
383 Set this bit to 1 to indicate that the value entered in the
384 heap_end_ptr is valid. If this field is clear, some setup code
385 functionality will be disabled.
387 Field name: setup_move_size
388 Type: modify (obligatory)
392 When using protocol 2.00 or 2.01, if the real mode kernel is not
393 loaded at 0x90000, it gets moved there later in the loading
394 sequence. Fill in this field if you want additional data (such as
395 the kernel command line) moved in addition to the real-mode kernel
398 The unit is bytes starting with the beginning of the boot sector.
400 This field is can be ignored when the protocol is 2.02 or higher, or
401 if the real-mode code is loaded at 0x90000.
403 Field name: code32_start
404 Type: modify (optional, reloc)
408 The address to jump to in protected mode. This defaults to the load
409 address of the kernel, and can be used by the boot loader to
410 determine the proper load address.
412 This field can be modified for two purposes:
414 1. as a boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
416 2. if a bootloader which does not install a hook loads a
417 relocatable kernel at a nonstandard address it will have to modify
418 this field to point to the load address.
420 Field name: ramdisk_image
421 Type: write (obligatory)
425 The 32-bit linear address of the initial ramdisk or ramfs. Leave at
426 zero if there is no initial ramdisk/ramfs.
428 Field name: ramdisk_size
429 Type: write (obligatory)
433 Size of the initial ramdisk or ramfs. Leave at zero if there is no
434 initial ramdisk/ramfs.
436 Field name: bootsect_kludge
437 Type: kernel internal
441 This field is obsolete.
443 Field name: heap_end_ptr
444 Type: write (obligatory)
448 Set this field to the offset (from the beginning of the real-mode
449 code) of the end of the setup stack/heap, minus 0x0200.
451 Field name: cmd_line_ptr
452 Type: write (obligatory)
456 Set this field to the linear address of the kernel command line.
457 The kernel command line can be located anywhere between the end of
458 the setup heap and 0xA0000; it does not have to be located in the
459 same 64K segment as the real-mode code itself.
461 Fill in this field even if your boot loader does not support a
462 command line, in which case you can point this to an empty string
463 (or better yet, to the string "auto".) If this field is left at
464 zero, the kernel will assume that your boot loader does not support
467 Field name: initrd_addr_max
472 The maximum address that may be occupied by the initial
473 ramdisk/ramfs contents. For boot protocols 2.02 or earlier, this
474 field is not present, and the maximum address is 0x37FFFFFF. (This
475 address is defined as the address of the highest safe byte, so if
476 your ramdisk is exactly 131072 bytes long and this field is
477 0x37FFFFFF, you can start your ramdisk at 0x37FE0000.)
479 Field name: kernel_alignment
484 Alignment unit required by the kernel (if relocatable_kernel is true.)
486 Field name: relocatable_kernel
491 If this field is nonzero, the protected-mode part of the kernel can
492 be loaded at any address that satisfies the kernel_alignment field.
493 After loading, the boot loader must set the code32_start field to
494 point to the loaded code, or to a boot loader hook.
496 Field name: cmdline_size
501 The maximum size of the command line without the terminating
502 zero. This means that the command line can contain at most
503 cmdline_size characters. With protocol version 2.05 and earlier, the
504 maximum size was 255.
506 Field name: hardware_subarch
511 In a paravirtualized environment the hardware low level architectural
512 pieces such as interrupt handling, page table handling, and
513 accessing process control registers needs to be done differently.
515 This field allows the bootloader to inform the kernel we are in one
516 one of those environments.
518 0x00000000 The default x86/PC environment
522 Field name: hardware_subarch_data
527 A pointer to data that is specific to hardware subarch
529 Field name: payload_offset
534 If non-zero then this field contains the offset from the end of the
535 real-mode code to the payload.
537 The payload may be compressed. The format of both the compressed and
538 uncompressed data should be determined using the standard magic
539 numbers. Currently only gzip compressed ELF is used.
541 Field name: payload_length
546 The length of the payload.
548 **** THE IMAGE CHECKSUM
550 From boot protocol version 2.08 onwards the CRC-32 is calculated over
551 the entire file using the characteristic polynomial 0x04C11DB7 and an
552 initial remainder of 0xffffffff. The checksum is appended to the
553 file; therefore the CRC of the file up to the limit specified in the
554 syssize field of the header is always 0.
556 **** THE KERNEL COMMAND LINE
558 The kernel command line has become an important way for the boot
559 loader to communicate with the kernel. Some of its options are also
560 relevant to the boot loader itself, see "special command line options"
563 The kernel command line is a null-terminated string. The maximum
564 length can be retrieved from the field cmdline_size. Before protocol
565 version 2.06, the maximum was 255 characters. A string that is too
566 long will be automatically truncated by the kernel.
568 If the boot protocol version is 2.02 or later, the address of the
569 kernel command line is given by the header field cmd_line_ptr (see
570 above.) This address can be anywhere between the end of the setup
573 If the protocol version is *not* 2.02 or higher, the kernel
574 command line is entered using the following protocol:
576 At offset 0x0020 (word), "cmd_line_magic", enter the magic
579 At offset 0x0022 (word), "cmd_line_offset", enter the offset
580 of the kernel command line (relative to the start of the
583 The kernel command line *must* be within the memory region
584 covered by setup_move_size, so you may need to adjust this
587 Field name: setup_data
588 Type: write (obligatory)
592 The 64-bit physical pointer to NULL terminated single linked list of
593 struct setup_data. This is used to define a more extensible boot
594 parameters passing mechanism. The definition of struct setup_data is
604 Where, the next is a 64-bit physical pointer to the next node of
605 linked list, the next field of the last node is 0; the type is used
606 to identify the contents of data; the len is the length of data
607 field; the data holds the real payload.
610 **** MEMORY LAYOUT OF THE REAL-MODE CODE
612 The real-mode code requires a stack/heap to be set up, as well as
613 memory allocated for the kernel command line. This needs to be done
614 in the real-mode accessible memory in bottom megabyte.
616 It should be noted that modern machines often have a sizable Extended
617 BIOS Data Area (EBDA). As a result, it is advisable to use as little
618 of the low megabyte as possible.
620 Unfortunately, under the following circumstances the 0x90000 memory
621 segment has to be used:
623 - When loading a zImage kernel ((loadflags & 0x01) == 0).
624 - When loading a 2.01 or earlier boot protocol kernel.
626 -> For the 2.00 and 2.01 boot protocols, the real-mode code
627 can be loaded at another address, but it is internally
628 relocated to 0x90000. For the "old" protocol, the
629 real-mode code must be loaded at 0x90000.
631 When loading at 0x90000, avoid using memory above 0x9a000.
633 For boot protocol 2.02 or higher, the command line does not have to be
634 located in the same 64K segment as the real-mode setup code; it is
635 thus permitted to give the stack/heap the full 64K segment and locate
636 the command line above it.
638 The kernel command line should not be located below the real-mode
639 code, nor should it be located in high memory.
642 **** SAMPLE BOOT CONFIGURATION
644 As a sample configuration, assume the following layout of the real
647 When loading below 0x90000, use the entire segment:
649 0x0000-0x7fff Real mode kernel
650 0x8000-0xdfff Stack and heap
651 0xe000-0xffff Kernel command line
653 When loading at 0x90000 OR the protocol version is 2.01 or earlier:
655 0x0000-0x7fff Real mode kernel
656 0x8000-0x97ff Stack and heap
657 0x9800-0x9fff Kernel command line
659 Such a boot loader should enter the following fields in the header:
661 unsigned long base_ptr; /* base address for real-mode segment */
663 if ( setup_sects == 0 ) {
667 if ( protocol >= 0x0200 ) {
668 type_of_loader = <type code>;
669 if ( loading_initrd ) {
670 ramdisk_image = <initrd_address>;
671 ramdisk_size = <initrd_size>;
674 if ( protocol >= 0x0202 && loadflags & 0x01 )
679 if ( protocol >= 0x0201 ) {
680 heap_end_ptr = heap_end - 0x200;
681 loadflags |= 0x80; /* CAN_USE_HEAP */
684 if ( protocol >= 0x0202 ) {
685 cmd_line_ptr = base_ptr + heap_end;
686 strcpy(cmd_line_ptr, cmdline);
688 cmd_line_magic = 0xA33F;
689 cmd_line_offset = heap_end;
690 setup_move_size = heap_end + strlen(cmdline)+1;
691 strcpy(base_ptr+cmd_line_offset, cmdline);
694 /* Very old kernel */
698 cmd_line_magic = 0xA33F;
699 cmd_line_offset = heap_end;
701 /* A very old kernel MUST have its real-mode code
704 if ( base_ptr != 0x90000 ) {
705 /* Copy the real-mode kernel */
706 memcpy(0x90000, base_ptr, (setup_sects+1)*512);
707 base_ptr = 0x90000; /* Relocated */
710 strcpy(0x90000+cmd_line_offset, cmdline);
712 /* It is recommended to clear memory up to the 32K mark */
713 memset(0x90000 + (setup_sects+1)*512, 0,
714 (64-(setup_sects+1))*512);
718 **** LOADING THE REST OF THE KERNEL
720 The 32-bit (non-real-mode) kernel starts at offset (setup_sects+1)*512
721 in the kernel file (again, if setup_sects == 0 the real value is 4.)
722 It should be loaded at address 0x10000 for Image/zImage kernels and
723 0x100000 for bzImage kernels.
725 The kernel is a bzImage kernel if the protocol >= 2.00 and the 0x01
726 bit (LOAD_HIGH) in the loadflags field is set:
728 is_bzImage = (protocol >= 0x0200) && (loadflags & 0x01);
729 load_address = is_bzImage ? 0x100000 : 0x10000;
731 Note that Image/zImage kernels can be up to 512K in size, and thus use
732 the entire 0x10000-0x90000 range of memory. This means it is pretty
733 much a requirement for these kernels to load the real-mode part at
734 0x90000. bzImage kernels allow much more flexibility.
737 **** SPECIAL COMMAND LINE OPTIONS
739 If the command line provided by the boot loader is entered by the
740 user, the user may expect the following command line options to work.
741 They should normally not be deleted from the kernel command line even
742 though not all of them are actually meaningful to the kernel. Boot
743 loader authors who need additional command line options for the boot
744 loader itself should get them registered in
745 Documentation/kernel-parameters.txt to make sure they will not
746 conflict with actual kernel options now or in the future.
749 <mode> here is either an integer (in C notation, either
750 decimal, octal, or hexadecimal) or one of the strings
751 "normal" (meaning 0xFFFF), "ext" (meaning 0xFFFE) or "ask"
752 (meaning 0xFFFD). This value should be entered into the
753 vid_mode field, as it is used by the kernel before the command
757 <size> is an integer in C notation optionally followed by
758 (case insensitive) K, M, G, T, P or E (meaning << 10, << 20,
759 << 30, << 40, << 50 or << 60). This specifies the end of
760 memory to the kernel. This affects the possible placement of
761 an initrd, since an initrd should be placed near end of
762 memory. Note that this is an option to *both* the kernel and
766 An initrd should be loaded. The meaning of <file> is
767 obviously bootloader-dependent, and some boot loaders
768 (e.g. LILO) do not have such a command.
770 In addition, some boot loaders add the following options to the
771 user-specified command line:
774 The boot image which was loaded. Again, the meaning of <file>
775 is obviously bootloader-dependent.
778 The kernel was booted without explicit user intervention.
780 If these options are added by the boot loader, it is highly
781 recommended that they are located *first*, before the user-specified
782 or configuration-specified command line. Otherwise, "init=/bin/sh"
783 gets confused by the "auto" option.
786 **** RUNNING THE KERNEL
788 The kernel is started by jumping to the kernel entry point, which is
789 located at *segment* offset 0x20 from the start of the real mode
790 kernel. This means that if you loaded your real-mode kernel code at
791 0x90000, the kernel entry point is 9020:0000.
793 At entry, ds = es = ss should point to the start of the real-mode
794 kernel code (0x9000 if the code is loaded at 0x90000), sp should be
795 set up properly, normally pointing to the top of the heap, and
796 interrupts should be disabled. Furthermore, to guard against bugs in
797 the kernel, it is recommended that the boot loader sets fs = gs = ds =
800 In our example from above, we would do:
802 /* Note: in the case of the "old" kernel protocol, base_ptr must
803 be == 0x90000 at this point; see the previous sample code */
807 cli(); /* Enter with interrupts disabled! */
809 /* Set up the real-mode kernel stack */
813 _DS = _ES = _FS = _GS = seg;
814 jmp_far(seg+0x20, 0); /* Run the kernel */
816 If your boot sector accesses a floppy drive, it is recommended to
817 switch off the floppy motor before running the kernel, since the
818 kernel boot leaves interrupts off and thus the motor will not be
819 switched off, especially if the loaded kernel has the floppy driver as
820 a demand-loaded module!
823 **** ADVANCED BOOT LOADER HOOKS
825 If the boot loader runs in a particularly hostile environment (such as
826 LOADLIN, which runs under DOS) it may be impossible to follow the
827 standard memory location requirements. Such a boot loader may use the
828 following hooks that, if set, are invoked by the kernel at the
829 appropriate time. The use of these hooks should probably be
830 considered an absolutely last resort!
832 IMPORTANT: All the hooks are required to preserve %esp, %ebp, %esi and
833 %edi across invocation.
836 A 16-bit real mode far subroutine invoked immediately before
837 entering protected mode. The default routine disables NMI, so
838 your routine should probably do so, too.
841 A 32-bit flat-mode routine *jumped* to immediately after the
842 transition to protected mode, but before the kernel is
843 uncompressed. No segments, except CS, are guaranteed to be
844 set up (current kernels do, but older ones do not); you should
845 set them up to BOOT_DS (0x18) yourself.
847 After completing your hook, you should jump to the address
848 that was in this field before your boot loader overwrote it
849 (relocated, if appropriate.)
852 **** 32-bit BOOT PROTOCOL
854 For machine with some new BIOS other than legacy BIOS, such as EFI,
855 LinuxBIOS, etc, and kexec, the 16-bit real mode setup code in kernel
856 based on legacy BIOS can not be used, so a 32-bit boot protocol needs
859 In 32-bit boot protocol, the first step in loading a Linux kernel
860 should be to setup the boot parameters (struct boot_params,
861 traditionally known as "zero page"). The memory for struct boot_params
862 should be allocated and initialized to all zero. Then the setup header
863 from offset 0x01f1 of kernel image on should be loaded into struct
864 boot_params and examined. The end of setup header can be calculated as
867 0x0202 + byte value at offset 0x0201
869 In addition to read/modify/write the setup header of the struct
870 boot_params as that of 16-bit boot protocol, the boot loader should
871 also fill the additional fields of the struct boot_params as that
872 described in zero-page.txt.
874 After setupping the struct boot_params, the boot loader can load the
875 32/64-bit kernel in the same way as that of 16-bit boot protocol.
877 In 32-bit boot protocol, the kernel is started by jumping to the
878 32-bit kernel entry point, which is the start address of loaded
881 At entry, the CPU must be in 32-bit protected mode with paging
882 disabled; a GDT must be loaded with the descriptors for selectors
883 __BOOT_CS(0x10) and __BOOT_DS(0x18); both descriptors must be 4G flat
884 segment; __BOOS_CS must have execute/read permission, and __BOOT_DS
885 must have read/write permission; CS must be __BOOT_CS and DS, ES, SS
886 must be __BOOT_DS; interrupt must be disabled; %esi must hold the base
887 address of the struct boot_params; %ebp, %edi and %ebx must be zero.