1 .. SPDX-License-Identifier: GPL-2.0
2
3 ===========================
4 The Linux/x86 Boot Protocol
5 ===========================
6
7 On the x86 platform, the Linux kernel uses a r
8 convention. This has evolved partially due to
9 well as the desire in the early days to have t
10 bootable image, the complicated PC memory mode
11 expectations in the PC industry caused by the
12 real-mode DOS as a mainstream operating system
13
14 Currently, the following versions of the Linux
15
16 ============= ==============================
17 Old kernels zImage/Image support only. So
18 may not even support a command
19
20 Protocol 2.00 (Kernel 1.3.73) Added bzImage
21 well as a formalized way to co
22 boot loader and the kernel. s
23 although the traditional setup
24 writable.
25
26 Protocol 2.01 (Kernel 1.3.76) Added a heap o
27
28 Protocol 2.02 (Kernel 2.4.0-test3-pre3) New
29 Lower the conventional memory
30 of the traditional setup area,
31 safe for systems which use the
32 BIOS entry points. zImage dep
33 supported.
34
35 Protocol 2.03 (Kernel 2.4.18-pre1) Explicitl
36 initrd address available to th
37
38 Protocol 2.04 (Kernel 2.6.14) Extend the sys
39
40 Protocol 2.05 (Kernel 2.6.20) Make protected
41 Introduce relocatable_kernel a
42
43 Protocol 2.06 (Kernel 2.6.22) Added a field
44 the boot command line.
45
46 Protocol 2.07 (Kernel 2.6.24) Added paravirt
47 Introduced hardware_subarch an
48 and KEEP_SEGMENTS flag in load
49
50 Protocol 2.08 (Kernel 2.6.26) Added crc32 ch
51 payload. Introduced payload_of
52 fields to aid in locating the
53
54 Protocol 2.09 (Kernel 2.6.26) Added a field
55 pointer to single linked list
56
57 Protocol 2.10 (Kernel 2.6.31) Added a protoc
58 beyond the kernel_alignment ad
59 pref_address fields. Added ex
60
61 Protocol 2.11 (Kernel 3.6) Added a field for
62 protocol entry point.
63
64 Protocol 2.12 (Kernel 3.8) Added the xloadfl
65 to struct boot_params for load
66 above 4G in 64bit.
67
68 Protocol 2.13 (Kernel 3.14) Support 32- and
69 xloadflags to support booting
70 EFI
71
72 Protocol 2.14 BURNT BY INCORRECT COMMIT
73 ae7e1238e68f2a472a125673ab506d
74 ("x86/boot: Add ACPI RSDP addr
75 DO NOT USE!!! ASSUME SAME AS 2
76
77 Protocol 2.15 (Kernel 5.5) Added the kernel_
78 ============= ==============================
79
80 .. note::
81 The protocol version number should be cha
82 is changed. There is no need to update th
83 or kernel_info are changed. Additionally,
84 xloadflags (in this case the protocol ver
85 updated either) or kernel_info to communi
86 features to the boot loader. Due to very
87 the original setup header every update to
88 with great care. Starting from the protoc
89 communicate things to the boot loader is
90
91
92 Memory Layout
93 =============
94
95 The traditional memory map for the kernel load
96 zImage kernels, typically looks like::
97
98 | |
99 0A0000 +------------------------+
100 | Reserved for BIOS |
101 09A000 +------------------------+
102 | Command line |
103 | Stack/heap |
104 098000 +------------------------+
105 | Kernel setup |
106 090200 +------------------------+
107 | Kernel boot sector |
108 090000 +------------------------+
109 | Protected-mode kernel |
110 010000 +------------------------+
111 | Boot loader |
112 001000 +------------------------+
113 | Reserved for MBR/BIOS |
114 000800 +------------------------+
115 | Typically used by MBR |
116 000600 +------------------------+
117 | BIOS use only |
118 000000 +------------------------+
119
120 When using bzImage, the protected-mode kernel
121 0x100000 ("high memory"), and the kernel real-
122 setup, and stack/heap) was made relocatable to
123 0x10000 and end of low memory. Unfortunately,
124 2.01 the 0x90000+ memory range is still used i
125 the 2.02 protocol resolves that problem.
126
127 It is desirable to keep the "memory ceiling" -
128 low memory touched by the boot loader -- as lo
129 some newer BIOSes have begun to allocate some
130 memory, called the Extended BIOS Data Area, ne
131 memory. The boot loader should use the "INT 1
132 how much low memory is available.
133
134 Unfortunately, if INT 12h reports that the amo
135 low, there is usually nothing the boot loader
136 error to the user. The boot loader should the
137 take up as little space in low memory as it re
138 zImage or old bzImage kernels, which need data
139 0x90000 segment, the boot loader should make s
140 above the 0x9A000 point; too many BIOSes will
141
142 For a modern bzImage kernel with boot protocol
143 memory layout like the following is suggested:
144
145 ~ ~
146 | Protected-mode kernel |
147 100000 +------------------------+
148 | I/O memory hole |
149 0A0000 +------------------------+
150 | Reserved for BIOS |
151 ~ ~
152 | Command line |
153 X+10000 +------------------------+
154 | Stack/heap |
155 X+08000 +------------------------+
156 | Kernel setup |
157 | Kernel boot sector |
158 X +------------------------+
159 | Boot loader |
160 001000 +------------------------+
161 | Reserved for MBR/BIOS |
162 000800 +------------------------+
163 | Typically used by MBR |
164 000600 +------------------------+
165 | BIOS use only |
166 000000 +------------------------+
167
168 ... where the address X is as low as the des
169
170
171 The Real-Mode Kernel Header
172 ===========================
173
174 In the following text, and anywhere in the ker
175 sector" refers to 512 bytes. It is independen
176 size of the underlying medium.
177
178 The first step in loading a Linux kernel shoul
179 real-mode code (boot sector and setup code) an
180 following header at offset 0x01f1. The real-m
181 32K, although the boot loader may choose to lo
182 sectors (1K) and then examine the bootup secto
183
184 The header looks like:
185
186 =========== ======== ==============
187 Offset/Size Proto Name
188 =========== ======== ==============
189 01F1/1 ALL(1) setup_sects
190 01F2/2 ALL root_flags
191 01F4/4 2.04+(2) syssize
192 01F8/2 ALL ram_size
193 01FA/2 ALL vid_mode
194 01FC/2 ALL root_dev
195 01FE/2 ALL boot_flag
196 0200/2 2.00+ jump
197 0202/4 2.00+ header
198 0206/2 2.00+ version
199 0208/4 2.00+ realmode_swtch
200 020C/2 2.00+ start_sys_seg
201 020E/2 2.00+ kernel_version
202 0210/1 2.00+ type_of_loader
203 0211/1 2.00+ loadflags
204 0212/2 2.00+ setup_move_siz
205 0214/4 2.00+ code32_start
206 0218/4 2.00+ ramdisk_image
207 021C/4 2.00+ ramdisk_size
208 0220/4 2.00+ bootsect_kludg
209 0224/2 2.01+ heap_end_ptr
210 0226/1 2.02+(3) ext_loader_ver
211 0227/1 2.02+(3) ext_loader_typ
212 0228/4 2.02+ cmd_line_ptr
213 022C/4 2.03+ initrd_addr_ma
214 0230/4 2.05+ kernel_alignme
215 0234/1 2.05+ relocatable_ke
216 0235/1 2.10+ min_alignment
217 0236/2 2.12+ xloadflags
218 0238/4 2.06+ cmdline_size
219 023C/4 2.07+ hardware_subar
220 0240/8 2.07+ hardware_subar
221 0248/4 2.08+ payload_offset
222 024C/4 2.08+ payload_length
223 0250/8 2.09+ setup_data
224
225 0258/8 2.10+ pref_address
226 0260/4 2.10+ init_size
227 0264/4 2.11+ handover_offse
228 0268/4 2.15+ kernel_info_of
229 =========== ======== ==============
230
231 .. note::
232 (1) For backwards compatibility, if the setu
233 real value is 4.
234
235 (2) For boot protocol prior to 2.04, the upp
236 field are unusable, which means the size
237 cannot be determined.
238
239 (3) Ignored, but safe to set, for boot proto
240
241 If the "HdrS" (0x53726448) magic number is not
242 the boot protocol version is "old". Loading a
243 following parameters should be assumed::
244
245 Image type = zImage
246 initrd not supported
247 Real-mode kernel must be located at 0x
248
249 Otherwise, the "version" field contains the pr
250 e.g. protocol version 2.01 will contain 0x0201
251 setting fields in the header, you must make su
252 supported by the protocol version in use.
253
254
255 Details of Header Fields
256 ========================
257
258 For each field, some are information from the
259 ("read"), some are expected to be filled out b
260 ("write"), and some are expected to be read an
261 bootloader ("modify").
262
263 All general purpose boot loaders should write
264 (obligatory). Boot loaders who want to load t
265 nonstandard address should fill in the fields
266 boot loaders can ignore those fields.
267
268 The byte order of all fields is littleendian (
269
270 ============ ===========
271 Field name: setup_sects
272 Type: read
273 Offset/size: 0x1f1/1
274 Protocol: ALL
275 ============ ===========
276
277 The size of the setup code in 512-byte secto
278 0, the real value is 4. The real-mode code
279 sector (always one 512-byte sector) plus the
280
281 ============ =================
282 Field name: root_flags
283 Type: modify (optional)
284 Offset/size: 0x1f2/2
285 Protocol: ALL
286 ============ =================
287
288 If this field is nonzero, the root defaults
289 this field is deprecated; use the "ro" or "r
290 command line instead.
291
292 ============ ==============================
293 Field name: syssize
294 Type: read
295 Offset/size: 0x1f4/4 (protocol 2.04+) 0x1f4
296 Protocol: 2.04+
297 ============ ==============================
298
299 The size of the protected-mode code in units
300 For protocol versions older than 2.04 this f
301 wide, and therefore cannot be trusted for th
302 the LOAD_HIGH flag is set.
303
304 ============ ===============
305 Field name: ram_size
306 Type: kernel internal
307 Offset/size: 0x1f8/2
308 Protocol: ALL
309 ============ ===============
310
311 This field is obsolete.
312
313 ============ ===================
314 Field name: vid_mode
315 Type: modify (obligatory)
316 Offset/size: 0x1fa/2
317 ============ ===================
318
319 Please see the section on SPECIAL COMMAND LI
320
321 ============ =================
322 Field name: root_dev
323 Type: modify (optional)
324 Offset/size: 0x1fc/2
325 Protocol: ALL
326 ============ =================
327
328 The default root device device number. The
329 deprecated, use the "root=" option on the co
330
331 ============ =========
332 Field name: boot_flag
333 Type: read
334 Offset/size: 0x1fe/2
335 Protocol: ALL
336 ============ =========
337
338 Contains 0xAA55. This is the closest thing
339 to a magic number.
340
341 ============ =======
342 Field name: jump
343 Type: read
344 Offset/size: 0x200/2
345 Protocol: 2.00+
346 ============ =======
347
348 Contains an x86 jump instruction, 0xEB follo
349 relative to byte 0x202. This can be used to
350 the header.
351
352 ============ =======
353 Field name: header
354 Type: read
355 Offset/size: 0x202/4
356 Protocol: 2.00+
357 ============ =======
358
359 Contains the magic number "HdrS" (0x53726448
360
361 ============ =======
362 Field name: version
363 Type: read
364 Offset/size: 0x206/2
365 Protocol: 2.00+
366 ============ =======
367
368 Contains the boot protocol version, in (majo
369 e.g. 0x0204 for version 2.04, and 0x0a11 for
370 10.17.
371
372 ============ =================
373 Field name: realmode_swtch
374 Type: modify (optional)
375 Offset/size: 0x208/4
376 Protocol: 2.00+
377 ============ =================
378
379 Boot loader hook (see ADVANCED BOOT LOADER H
380
381 ============ =============
382 Field name: start_sys_seg
383 Type: read
384 Offset/size: 0x20c/2
385 Protocol: 2.00+
386 ============ =============
387
388 The load low segment (0x1000). Obsolete.
389
390 ============ ==============
391 Field name: kernel_version
392 Type: read
393 Offset/size: 0x20e/2
394 Protocol: 2.00+
395 ============ ==============
396
397 If set to a nonzero value, contains a pointe
398 human-readable kernel version number string,
399 be used to display the kernel version to the
400 should be less than (0x200*setup_sects).
401
402 For example, if this value is set to 0x1c00,
403 number string can be found at offset 0x1e00
404 This is a valid value if and only if the "se
405 contains the value 15 or higher, as::
406
407 0x1c00 < 15*0x200 (= 0x1e00) but
408 0x1c00 >= 14*0x200 (= 0x1c00)
409
410 0x1c00 >> 9 = 14, So the minimum value
411
412 ============ ==================
413 Field name: type_of_loader
414 Type: write (obligatory)
415 Offset/size: 0x210/1
416 Protocol: 2.00+
417 ============ ==================
418
419 If your boot loader has an assigned id (see
420 0xTV here, where T is an identifier for the
421 a version number. Otherwise, enter 0xFF her
422
423 For boot loader IDs above T = 0xD, write T =
424 write the extended ID minus 0x10 to the ext_
425 Similarly, the ext_loader_ver field can be u
426 four bits for the bootloader version.
427
428 For example, for T = 0x15, V = 0x234, write:
429
430 type_of_loader <- 0xE4
431 ext_loader_type <- 0x05
432 ext_loader_ver <- 0x23
433
434 Assigned boot loader ids (hexadecimal):
435
436 == ===================================
437 0 LILO
438 (0x00 reserved for pre-2.00 bootloa
439 1 Loadlin
440 2 bootsect-loader
441 (0x20, all other values reserved)
442 3 Syslinux
443 4 Etherboot/gPXE/iPXE
444 5 ELILO
445 7 GRUB
446 8 U-Boot
447 9 Xen
448 A Gujin
449 B Qemu
450 C Arcturus Networks uCbootloader
451 D kexec-tools
452 E Extended (see ext_loader_type)
453 F Special (0xFF = undefined)
454 10 Reserved
455 11 Minimal Linux Bootloader
456 <http://sebastian-plotz.blogspot.de
457 12 OVMF UEFI virtualization stack
458 13 barebox
459 == ===================================
460
461 Please contact <hpa@zytor.com> if you need a
462
463 ============ ===================
464 Field name: loadflags
465 Type: modify (obligatory)
466 Offset/size: 0x211/1
467 Protocol: 2.00+
468 ============ ===================
469
470 This field is a bitmask.
471
472 Bit 0 (read): LOADED_HIGH
473
474 - If 0, the protected-mode code is loa
475 - If 1, the protected-mode code is loa
476
477 Bit 1 (kernel internal): KASLR_FLAG
478
479 - Used internally by the compressed ke
480 KASLR status to kernel proper.
481
482 - If 1, KASLR enabled.
483 - If 0, KASLR disabled.
484
485 Bit 5 (write): QUIET_FLAG
486
487 - If 0, print early messages.
488 - If 1, suppress early messages.
489
490 This requests to the kernel (d
491 kernel) to not write early mes
492 accessing the display hardware
493
494 Bit 6 (obsolete): KEEP_SEGMENTS
495
496 Protocol: 2.07+
497
498 - This flag is obsolete.
499
500 Bit 7 (write): CAN_USE_HEAP
501
502 Set this bit to 1 to indicate that the
503 heap_end_ptr is valid. If this field
504 functionality will be disabled.
505
506
507 ============ ===================
508 Field name: setup_move_size
509 Type: modify (obligatory)
510 Offset/size: 0x212/2
511 Protocol: 2.00-2.01
512 ============ ===================
513
514 When using protocol 2.00 or 2.01, if the rea
515 loaded at 0x90000, it gets moved there later
516 sequence. Fill in this field if you want ad
517 the kernel command line) moved in addition t
518 itself.
519
520 The unit is bytes starting with the beginnin
521
522 This field is can be ignored when the protoc
523 if the real-mode code is loaded at 0x90000.
524
525 ============ ========================
526 Field name: code32_start
527 Type: modify (optional, reloc)
528 Offset/size: 0x214/4
529 Protocol: 2.00+
530 ============ ========================
531
532 The address to jump to in protected mode. T
533 address of the kernel, and can be used by th
534 determine the proper load address.
535
536 This field can be modified for two purposes:
537
538 1. as a boot loader hook (see Advanced Boo
539
540 2. if a bootloader which does not install
541 relocatable kernel at a nonstandard add
542 this field to point to the load address
543
544 ============ ==================
545 Field name: ramdisk_image
546 Type: write (obligatory)
547 Offset/size: 0x218/4
548 Protocol: 2.00+
549 ============ ==================
550
551 The 32-bit linear address of the initial ram
552 zero if there is no initial ramdisk/ramfs.
553
554 ============ ==================
555 Field name: ramdisk_size
556 Type: write (obligatory)
557 Offset/size: 0x21c/4
558 Protocol: 2.00+
559 ============ ==================
560
561 Size of the initial ramdisk or ramfs. Leave
562 initial ramdisk/ramfs.
563
564 ============ ===============
565 Field name: bootsect_kludge
566 Type: kernel internal
567 Offset/size: 0x220/4
568 Protocol: 2.00+
569 ============ ===============
570
571 This field is obsolete.
572
573 ============ ==================
574 Field name: heap_end_ptr
575 Type: write (obligatory)
576 Offset/size: 0x224/2
577 Protocol: 2.01+
578 ============ ==================
579
580 Set this field to the offset (from the begin
581 code) of the end of the setup stack/heap, mi
582
583 ============ ================
584 Field name: ext_loader_ver
585 Type: write (optional)
586 Offset/size: 0x226/1
587 Protocol: 2.02+
588 ============ ================
589
590 This field is used as an extension of the ve
591 type_of_loader field. The total version num
592 (type_of_loader & 0x0f) + (ext_loader_ver <<
593
594 The use of this field is boot loader specifi
595 is zero.
596
597 Kernels prior to 2.6.31 did not recognize th
598 to write for protocol version 2.02 or higher
599
600 ============ ==============================
601 Field name: ext_loader_type
602 Type: write (obligatory if (type_of_
603 Offset/size: 0x227/1
604 Protocol: 2.02+
605 ============ ==============================
606
607 This field is used as an extension of the ty
608 type_of_loader field. If the type in type_o
609 the actual type is (ext_loader_type + 0x10).
610
611 This field is ignored if the type in type_of
612
613 Kernels prior to 2.6.31 did not recognize th
614 to write for protocol version 2.02 or higher
615
616 ============ ==================
617 Field name: cmd_line_ptr
618 Type: write (obligatory)
619 Offset/size: 0x228/4
620 Protocol: 2.02+
621 ============ ==================
622
623 Set this field to the linear address of the
624 The kernel command line can be located anywh
625 the setup heap and 0xA0000; it does not have
626 same 64K segment as the real-mode code itsel
627
628 Fill in this field even if your boot loader
629 command line, in which case you can point th
630 (or better yet, to the string "auto".) If t
631 zero, the kernel will assume that your boot
632 the 2.02+ protocol.
633
634 ============ ===============
635 Field name: initrd_addr_max
636 Type: read
637 Offset/size: 0x22c/4
638 Protocol: 2.03+
639 ============ ===============
640
641 The maximum address that may be occupied by
642 ramdisk/ramfs contents. For boot protocols
643 field is not present, and the maximum addres
644 address is defined as the address of the hig
645 your ramdisk is exactly 131072 bytes long an
646 0x37FFFFFF, you can start your ramdisk at 0x
647
648 ============ ============================
649 Field name: kernel_alignment
650 Type: read/modify (reloc)
651 Offset/size: 0x230/4
652 Protocol: 2.05+ (read), 2.10+ (modify)
653 ============ ============================
654
655 Alignment unit required by the kernel (if re
656 true.) A relocatable kernel that is loaded
657 incompatible with the value in this field wi
658 kernel initialization.
659
660 Starting with protocol version 2.10, this re
661 alignment preferred for optimal performance;
662 loader to modify this field to permit a less
663 min_alignment and pref_address field below.
664
665 ============ ==================
666 Field name: relocatable_kernel
667 Type: read (reloc)
668 Offset/size: 0x234/1
669 Protocol: 2.05+
670 ============ ==================
671
672 If this field is nonzero, the protected-mode
673 be loaded at any address that satisfies the
674 After loading, the boot loader must set the
675 point to the loaded code, or to a boot loade
676
677 ============ =============
678 Field name: min_alignment
679 Type: read (reloc)
680 Offset/size: 0x235/1
681 Protocol: 2.10+
682 ============ =============
683
684 This field, if nonzero, indicates as a power
685 alignment required, as opposed to preferred,
686 If a boot loader makes use of this field, it
687 kernel_alignment field with the alignment un
688
689 kernel_alignment = 1 << min_alignment
690
691 There may be a considerable performance cost
692 misaligned kernel. Therefore, a loader shou
693 power-of-two alignment from kernel_alignment
694
695 ============ ==========
696 Field name: xloadflags
697 Type: read
698 Offset/size: 0x236/2
699 Protocol: 2.12+
700 ============ ==========
701
702 This field is a bitmask.
703
704 Bit 0 (read): XLF_KERNEL_64
705
706 - If 1, this kernel has the legacy 64-
707
708 Bit 1 (read): XLF_CAN_BE_LOADED_ABOVE_4G
709
710 - If 1, kernel/boot_params/cmdline/ram
711
712 Bit 2 (read): XLF_EFI_HANDOVER_32
713
714 - If 1, the kernel supports the 32-bit
715 given at handover_offset.
716
717 Bit 3 (read): XLF_EFI_HANDOVER_64
718
719 - If 1, the kernel supports the 64-bit
720 given at handover_offset + 0x200.
721
722 Bit 4 (read): XLF_EFI_KEXEC
723
724 - If 1, the kernel supports kexec EFI
725
726
727 ============ ============
728 Field name: cmdline_size
729 Type: read
730 Offset/size: 0x238/4
731 Protocol: 2.06+
732 ============ ============
733
734 The maximum size of the command line without
735 zero. This means that the command line can c
736 cmdline_size characters. With protocol versi
737 maximum size was 255.
738
739 ============ ==============================
740 Field name: hardware_subarch
741 Type: write (optional, defaults to x
742 Offset/size: 0x23c/4
743 Protocol: 2.07+
744 ============ ==============================
745
746 In a paravirtualized environment the hardwar
747 pieces such as interrupt handling, page tabl
748 accessing process control registers needs to
749
750 This field allows the bootloader to inform t
751 one of those environments.
752
753 ========== ==============================
754 0x00000000 The default x86/PC environment
755 0x00000001 lguest
756 0x00000002 Xen
757 0x00000003 Moorestown MID
758 0x00000004 CE4100 TV Platform
759 ========== ==============================
760
761 ============ =========================
762 Field name: hardware_subarch_data
763 Type: write (subarch-dependent)
764 Offset/size: 0x240/8
765 Protocol: 2.07+
766 ============ =========================
767
768 A pointer to data that is specific to hardwa
769 This field is currently unused for the defau
770 do not modify.
771
772 ============ ==============
773 Field name: payload_offset
774 Type: read
775 Offset/size: 0x248/4
776 Protocol: 2.08+
777 ============ ==============
778
779 If non-zero then this field contains the off
780 of the protected-mode code to the payload.
781
782 The payload may be compressed. The format of
783 uncompressed data should be determined using
784 numbers. The currently supported compressio
785 (magic numbers 1F 8B or 1F 9E), bzip2 (magic
786 (magic number 5D 00), XZ (magic number FD 37
787 02 21) and ZSTD (magic number 28 B5). The un
788 currently always ELF (magic number 7F 45 4C
789
790 ============ ==============
791 Field name: payload_length
792 Type: read
793 Offset/size: 0x24c/4
794 Protocol: 2.08+
795 ============ ==============
796
797 The length of the payload.
798
799 ============ ===============
800 Field name: setup_data
801 Type: write (special)
802 Offset/size: 0x250/8
803 Protocol: 2.09+
804 ============ ===============
805
806 The 64-bit physical pointer to NULL terminat
807 struct setup_data. This is used to define a
808 parameters passing mechanism. The definition
809 as follow::
810
811 struct setup_data {
812 u64 next;
813 u32 type;
814 u32 len;
815 u8 data[0];
816 };
817
818 Where, the next is a 64-bit physical pointer
819 linked list, the next field of the last node
820 to identify the contents of data; the len is
821 field; the data holds the real payload.
822
823 This list may be modified at a number of poi
824 process. Therefore, when modifying this lis
825 sure to consider the case where the linked l
826 entries.
827
828 The setup_data is a bit awkward to use for e
829 both because the setup_data header has to be
830 and because it has a 32-bit length field. Ho
831 intermediate stages of the boot process have
832 chunks of memory are occupied by kernel data
833
834 Thus setup_indirect struct and SETUP_INDIREC
835 protocol 2.15::
836
837 struct setup_indirect {
838 __u32 type;
839 __u32 reserved; /* Reserved, must be se
840 __u64 len;
841 __u64 addr;
842 };
843
844 The type member is a SETUP_INDIRECT | SETUP_
845 SETUP_INDIRECT itself since making the setup
846 could require a lot of stack space in someth
847 and stack space can be limited in boot conte
848
849 Let's give an example how to point to SETUP_
850 In this case setup_data and setup_indirect w
851
852 struct setup_data {
853 __u64 next = 0 or <addr_of_next_setup_da
854 __u32 type = SETUP_INDIRECT;
855 __u32 len = sizeof(setup_indirect);
856 __u8 data[sizeof(setup_indirect)] = stru
857 __u32 type = SETUP_INDIRECT | SETUP_E8
858 __u32 reserved = 0;
859 __u64 len = <len_of_SETUP_E820_EXT_dat
860 __u64 addr = <addr_of_SETUP_E820_EXT_d
861 }
862 }
863
864 .. note::
865 SETUP_INDIRECT | SETUP_NONE objects canno
866 from SETUP_INDIRECT itself. So, this kind
867 by the bootloaders.
868
869 ============ ============
870 Field name: pref_address
871 Type: read (reloc)
872 Offset/size: 0x258/8
873 Protocol: 2.10+
874 ============ ============
875
876 This field, if nonzero, represents a preferr
877 kernel. A relocating bootloader should atte
878 address if possible.
879
880 A non-relocatable kernel will unconditionall
881 at this address. A relocatable kernel will m
882 loaded below this address.
883
884 ============ =======
885 Field name: init_size
886 Type: read
887 Offset/size: 0x260/4
888 ============ =======
889
890 This field indicates the amount of linear co
891 at the kernel runtime start address that the
892 is capable of examining its memory map. Thi
893 as the total amount of memory the kernel nee
894 be used by a relocating boot loader to help
895 address for the kernel.
896
897 The kernel runtime start address is determin
898
899 if (relocatable_kernel)
900 runtime_start = align_up(load_address,
901 else
902 runtime_start = pref_address
903
904 ============ ===============
905 Field name: handover_offset
906 Type: read
907 Offset/size: 0x264/4
908 ============ ===============
909
910 This field is the offset from the beginning
911 the EFI handover protocol entry point. Boot
912 handover protocol to boot the kernel should
913
914 See EFI HANDOVER PROTOCOL below for more det
915
916 ============ ==================
917 Field name: kernel_info_offset
918 Type: read
919 Offset/size: 0x268/4
920 Protocol: 2.15+
921 ============ ==================
922
923 This field is the offset from the beginning
924 kernel_info. The kernel_info structure is em
925 in the uncompressed protected mode region.
926
927
928 The kernel_info
929 ===============
930
931 The relationships between the headers are anal
932 sections:
933
934 setup_header = .data
935 boot_params/setup_data = .bss
936
937 What is missing from the above list? That's ri
938
939 kernel_info = .rodata
940
941 We have been (ab)using .data for things that c
942 a long time, for lack of alternatives and -- e
943 Also, the BIOS stub is responsible for creatin
944 available to a BIOS-based loader (setup_data i
945
946 setup_header is permanently limited to 144 byt
947 2-byte jump field, which doubles as a length f
948 with the size of the "hole" in struct boot_par
949 or the BIOS stub has to copy it into. It is cu
950 leaves us with 25 very precious bytes. This is
951 without revising the boot protocol entirely, b
952
953 boot_params proper is limited to 4096 bytes, b
954 by adding setup_data entries. It cannot be use
955 the kernel image, because it is .bss and has n
956
957 kernel_info solves this by providing an extens
958 the kernel image. It is readonly, because the
959 bootloader copying its contents anywhere, but
960 necessary it can still contain data items that
961 expected to copy into a setup_data chunk.
962
963 All kernel_info data should be part of this st
964 be put before kernel_info_var_len_data label.
965 after kernel_info_var_len_data label. Each chu
966 be prefixed with header/magic and its size, e.
967
968 kernel_info:
969 .ascii "LToP" /* Header, L
970 .long kernel_info_var_len_data - k
971 .long kernel_info_end - kernel_inf
972 .long 0x01234567 /* Some fixe
973 kernel_info_var_len_data:
974 example_struct: /* Some vari
975 .ascii "0123" /* Header/Ma
976 .long example_struct_end - example
977 .ascii "Struct"
978 .long 0x89012345
979 example_struct_end:
980 example_strings: /* Some vari
981 .ascii "ABCD" /* Header/Ma
982 .long example_strings_end - exampl
983 .asciz "String_0"
984 .asciz "String_1"
985 example_strings_end:
986 kernel_info_end:
987
988 This way the kernel_info is self-contained blo
989
990 .. note::
991 Each variable size data header/magic can
992 without \0 at the end of the string, whic
993 existing variable length data headers/mag
994
995
996 Details of the kernel_info Fields
997 =================================
998
999 ============ ========
1000 Field name: header
1001 Offset/size: 0x0000/4
1002 ============ ========
1003
1004 Contains the magic number "LToP" (0x506f544
1005
1006 ============ ========
1007 Field name: size
1008 Offset/size: 0x0004/4
1009 ============ ========
1010
1011 This field contains the size of the kernel_
1012 It does not count kernel_info.kernel_info_v
1013 used by the bootloaders to detect supported
1014 and beginning of kernel_info.kernel_info_va
1015
1016 ============ ========
1017 Field name: size_total
1018 Offset/size: 0x0008/4
1019 ============ ========
1020
1021 This field contains the size of the kernel_
1022 and kernel_info.kernel_info_var_len_data.
1023
1024 ============ ==============
1025 Field name: setup_type_max
1026 Offset/size: 0x000c/4
1027 ============ ==============
1028
1029 This field contains maximal allowed type fo
1030
1031
1032 The Image Checksum
1033 ==================
1034
1035 From boot protocol version 2.08 onwards the C
1036 the entire file using the characteristic poly
1037 initial remainder of 0xffffffff. The checksu
1038 file; therefore the CRC of the file up to the
1039 syssize field of the header is always 0.
1040
1041
1042 The Kernel Command Line
1043 =======================
1044
1045 The kernel command line has become an importa
1046 loader to communicate with the kernel. Some
1047 relevant to the boot loader itself, see "spec
1048 below.
1049
1050 The kernel command line is a null-terminated
1051 length can be retrieved from the field cmdlin
1052 version 2.06, the maximum was 255 characters.
1053 long will be automatically truncated by the k
1054
1055 If the boot protocol version is 2.02 or later
1056 kernel command line is given by the header fi
1057 above.) This address can be anywhere between
1058 heap and 0xA0000.
1059
1060 If the protocol version is *not* 2.02 or high
1061 command line is entered using the following p
1062
1063 - At offset 0x0020 (word), "cmd_line_magic"
1064 number 0xA33F.
1065
1066 - At offset 0x0022 (word), "cmd_line_offset
1067 of the kernel command line (relative to t
1068 real-mode kernel).
1069
1070 - The kernel command line *must* be within
1071 covered by setup_move_size, so you may ne
1072 field.
1073
1074
1075 Memory Layout of The Real-Mode Code
1076 ===================================
1077
1078 The real-mode code requires a stack/heap to b
1079 memory allocated for the kernel command line.
1080 in the real-mode accessible memory in bottom
1081
1082 It should be noted that modern machines often
1083 BIOS Data Area (EBDA). As a result, it is ad
1084 of the low megabyte as possible.
1085
1086 Unfortunately, under the following circumstan
1087 segment has to be used:
1088
1089 - When loading a zImage kernel ((load
1090 - When loading a 2.01 or earlier boot
1091
1092 .. note::
1093 For the 2.00 and 2.01 boot protocols, th
1094 can be loaded at another address, but it
1095 relocated to 0x90000. For the "old" pro
1096 real-mode code must be loaded at 0x90000
1097
1098 When loading at 0x90000, avoid using memory a
1099
1100 For boot protocol 2.02 or higher, the command
1101 located in the same 64K segment as the real-m
1102 thus permitted to give the stack/heap the ful
1103 the command line above it.
1104
1105 The kernel command line should not be located
1106 code, nor should it be located in high memory
1107
1108
1109 Sample Boot Configuration
1110 =========================
1111
1112 As a sample configuration, assume the followi
1113 mode segment.
1114
1115 When loading below 0x90000, use the entir
1116
1117 ============= ===================
1118 0x0000-0x7fff Real mode kernel
1119 0x8000-0xdfff Stack and heap
1120 0xe000-0xffff Kernel command line
1121 ============= ===================
1122
1123 When loading at 0x90000 OR the protocol v
1124
1125 ============= ===================
1126 0x0000-0x7fff Real mode kernel
1127 0x8000-0x97ff Stack and heap
1128 0x9800-0x9fff Kernel command line
1129 ============= ===================
1130
1131 Such a boot loader should enter the following
1132
1133 unsigned long base_ptr; /* base addre
1134
1135 if ( setup_sects == 0 ) {
1136 setup_sects = 4;
1137 }
1138
1139 if ( protocol >= 0x0200 ) {
1140 type_of_loader = <type code>;
1141 if ( loading_initrd ) {
1142 ramdisk_image = <init
1143 ramdisk_size = <initr
1144 }
1145
1146 if ( protocol >= 0x0202 && lo
1147 heap_end = 0xe000;
1148 else
1149 heap_end = 0x9800;
1150
1151 if ( protocol >= 0x0201 ) {
1152 heap_end_ptr = heap_e
1153 loadflags |= 0x80; /*
1154 }
1155
1156 if ( protocol >= 0x0202 ) {
1157 cmd_line_ptr = base_p
1158 strcpy(cmd_line_ptr,
1159 } else {
1160 cmd_line_magic = 0xA
1161 cmd_line_offset = hea
1162 setup_move_size = hea
1163 strcpy(base_ptr+cmd_l
1164 }
1165 } else {
1166 /* Very old kernel */
1167
1168 heap_end = 0x9800;
1169
1170 cmd_line_magic = 0xA33F;
1171 cmd_line_offset = heap_end;
1172
1173 /* A very old kernel MUST hav
1174 loaded at 0x90000 */
1175
1176 if ( base_ptr != 0x90000 ) {
1177 /* Copy the real-mode
1178 memcpy(0x90000, base_
1179 base_ptr = 0x90000;
1180 }
1181
1182 strcpy(0x90000+cmd_line_offse
1183
1184 /* It is recommended to clear
1185 memset(0x90000 + (setup_sects
1186 (64-(setup_sects+1))*5
1187 }
1188
1189
1190 Loading The Rest of The Kernel
1191 ==============================
1192
1193 The 32-bit (non-real-mode) kernel starts at o
1194 in the kernel file (again, if setup_sects ==
1195 It should be loaded at address 0x10000 for Im
1196 0x100000 for bzImage kernels.
1197
1198 The kernel is a bzImage kernel if the protoco
1199 bit (LOAD_HIGH) in the loadflags field is set
1200
1201 is_bzImage = (protocol >= 0x0200) &&
1202 load_address = is_bzImage ? 0x100000
1203
1204 Note that Image/zImage kernels can be up to 5
1205 the entire 0x10000-0x90000 range of memory.
1206 much a requirement for these kernels to load
1207 0x90000. bzImage kernels allow much more fle
1208
1209 Special Command Line Options
1210 ============================
1211
1212 If the command line provided by the boot load
1213 user, the user may expect the following comma
1214 They should normally not be deleted from the
1215 though not all of them are actually meaningfu
1216 loader authors who need additional command li
1217 loader itself should get them registered in
1218 Documentation/admin-guide/kernel-parameters.r
1219 conflict with actual kernel options now or in
1220
1221 vga=<mode>
1222 <mode> here is either an integer (in
1223 decimal, octal, or hexadecimal) or on
1224 "normal" (meaning 0xFFFF), "ext" (mea
1225 (meaning 0xFFFD). This value should
1226 vid_mode field, as it is used by the
1227 line is parsed.
1228
1229 mem=<size>
1230 <size> is an integer in C notation op
1231 (case insensitive) K, M, G, T, P or E
1232 << 30, << 40, << 50 or << 60). This
1233 memory to the kernel. This affects th
1234 an initrd, since an initrd should be
1235 memory. Note that this is an option
1236 the bootloader!
1237
1238 initrd=<file>
1239 An initrd should be loaded. The mean
1240 obviously bootloader-dependent, and s
1241 (e.g. LILO) do not have such a comman
1242
1243 In addition, some boot loaders add the follow
1244 user-specified command line:
1245
1246 BOOT_IMAGE=<file>
1247 The boot image which was loaded. Aga
1248 is obviously bootloader-dependent.
1249
1250 auto
1251 The kernel was booted without explici
1252
1253 If these options are added by the boot loader
1254 recommended that they are located *first*, be
1255 or configuration-specified command line. Oth
1256 gets confused by the "auto" option.
1257
1258
1259 Running the Kernel
1260 ==================
1261
1262 The kernel is started by jumping to the kerne
1263 located at *segment* offset 0x20 from the sta
1264 kernel. This means that if you loaded your r
1265 0x90000, the kernel entry point is 9020:0000.
1266
1267 At entry, ds = es = ss should point to the st
1268 kernel code (0x9000 if the code is loaded at
1269 set up properly, normally pointing to the top
1270 interrupts should be disabled. Furthermore,
1271 the kernel, it is recommended that the boot l
1272 es = ss.
1273
1274 In our example from above, we would do::
1275
1276 /* Note: in the case of the "old" ker
1277 be == 0x90000 at this point; see t
1278
1279 seg = base_ptr >> 4;
1280
1281 cli(); /* Enter with interrupts disa
1282
1283 /* Set up the real-mode kernel stack
1284 _SS = seg;
1285 _SP = heap_end;
1286
1287 _DS = _ES = _FS = _GS = seg;
1288 jmp_far(seg+0x20, 0); /* Run the ke
1289
1290 If your boot sector accesses a floppy drive,
1291 switch off the floppy motor before running th
1292 kernel boot leaves interrupts off and thus th
1293 switched off, especially if the loaded kernel
1294 a demand-loaded module!
1295
1296
1297 Advanced Boot Loader Hooks
1298 ==========================
1299
1300 If the boot loader runs in a particularly hos
1301 LOADLIN, which runs under DOS) it may be impo
1302 standard memory location requirements. Such
1303 following hooks that, if set, are invoked by
1304 appropriate time. The use of these hooks sho
1305 considered an absolutely last resort!
1306
1307 IMPORTANT: All the hooks are required to pres
1308 %edi across invocation.
1309
1310 realmode_swtch:
1311 A 16-bit real mode far subroutine inv
1312 entering protected mode. The default
1313 your routine should probably do so, t
1314
1315 code32_start:
1316 A 32-bit flat-mode routine *jumped* t
1317 transition to protected mode, but bef
1318 uncompressed. No segments, except CS
1319 set up (current kernels do, but older
1320 set them up to BOOT_DS (0x18) yoursel
1321
1322 After completing your hook, you shoul
1323 that was in this field before your bo
1324 (relocated, if appropriate.)
1325
1326
1327 32-bit Boot Protocol
1328 ====================
1329
1330 For machine with some new BIOS other than leg
1331 LinuxBIOS, etc, and kexec, the 16-bit real mo
1332 based on legacy BIOS can not be used, so a 32
1333 to be defined.
1334
1335 In 32-bit boot protocol, the first step in lo
1336 should be to setup the boot parameters (struc
1337 traditionally known as "zero page"). The memo
1338 should be allocated and initialized to all ze
1339 from offset 0x01f1 of kernel image on should
1340 boot_params and examined. The end of setup he
1341 follow::
1342
1343 0x0202 + byte value at offset 0x0201
1344
1345 In addition to read/modify/write the setup he
1346 boot_params as that of 16-bit boot protocol,
1347 also fill the additional fields of the struct
1348 described in chapter Documentation/arch/x86/z
1349
1350 After setting up the struct boot_params, the
1351 32/64-bit kernel in the same way as that of 1
1352
1353 In 32-bit boot protocol, the kernel is starte
1354 32-bit kernel entry point, which is the start
1355 32/64-bit kernel.
1356
1357 At entry, the CPU must be in 32-bit protected
1358 disabled; a GDT must be loaded with the descr
1359 __BOOT_CS(0x10) and __BOOT_DS(0x18); both des
1360 segment; __BOOT_CS must have execute/read per
1361 must have read/write permission; CS must be _
1362 must be __BOOT_DS; interrupt must be disabled
1363 address of the struct boot_params; %ebp, %edi
1364
1365 64-bit Boot Protocol
1366 ====================
1367
1368 For machine with 64bit cpus and 64bit kernel,
1369 and we need a 64-bit boot protocol.
1370
1371 In 64-bit boot protocol, the first step in lo
1372 should be to setup the boot parameters (struc
1373 traditionally known as "zero page"). The memo
1374 could be allocated anywhere (even above 4G) a
1375 Then, the setup header at offset 0x01f1 of ke
1376 loaded into struct boot_params and examined.
1377 can be calculated as follows::
1378
1379 0x0202 + byte value at offset 0x0201
1380
1381 In addition to read/modify/write the setup he
1382 boot_params as that of 16-bit boot protocol,
1383 also fill the additional fields of the struct
1384 in chapter Documentation/arch/x86/zero-page.r
1385
1386 After setting up the struct boot_params, the
1387 64-bit kernel in the same way as that of 16-b
1388 kernel could be loaded above 4G.
1389
1390 In 64-bit boot protocol, the kernel is starte
1391 64-bit kernel entry point, which is the start
1392 64-bit kernel plus 0x200.
1393
1394 At entry, the CPU must be in 64-bit mode with
1395 The range with setup_header.init_size from st
1396 kernel and zero page and command line buffer
1397 a GDT must be loaded with the descriptors for
1398 __BOOT_CS(0x10) and __BOOT_DS(0x18); both des
1399 segment; __BOOT_CS must have execute/read per
1400 must have read/write permission; CS must be _
1401 must be __BOOT_DS; interrupt must be disabled
1402 address of the struct boot_params.
1403
1404 EFI Handover Protocol (deprecated)
1405 ==================================
1406
1407 This protocol allows boot loaders to defer in
1408 boot stub. The boot loader is required to loa
1409 from the boot media and jump to the EFI hando
1410 which is hdr->handover_offset bytes from the
1411 startup_{32,64}.
1412
1413 The boot loader MUST respect the kernel's PE/
1414 to section alignment, the memory footprint of
1415 the size of the file itself, and any other as
1416 that may affect correct operation of the imag
1417 execution context provided by the EFI firmwar
1418
1419 The function prototype for the handover entry
1420
1421 efi_stub_entry(void *handle, efi_system_t
1422
1423 'handle' is the EFI image handle passed to th
1424 firmware, 'table' is the EFI system table - t
1425 arguments of the "handoff state" as described
1426 UEFI specification. 'bp' is the boot loader-a
1427
1428 The boot loader *must* fill out the following
1429
1430 - hdr.cmd_line_ptr
1431 - hdr.ramdisk_image (if applicable)
1432 - hdr.ramdisk_size (if applicable)
1433
1434 All other fields should be zero.
1435
1436 NOTE: The EFI Handover Protocol is deprecated
1437 entry point, combined with the LINUX_EF
1438 loading protocol (refer to [0] for an e
1439 this), which removes the need for any k
1440 bootloader regarding the internal repre
1441 requirements/limitations regarding the
1442 and ramdisk in memory, or the placement
1443
1444 [0] https://github.com/u-boot/u-boot/commit/e
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.