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, four 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
48 The traditional memory map for the kernel loader, used for Image or
49 zImage kernels, typically looks like:
52 0A0000 +------------------------+
53 | Reserved for BIOS | Do not use. Reserved for BIOS EBDA.
54 09A000 +------------------------+
55 | Stack/heap/cmdline | For use by the kernel real-mode code.
56 098000 +------------------------+
57 | Kernel setup | The kernel real-mode code.
58 090200 +------------------------+
59 | Kernel boot sector | The kernel legacy boot sector.
60 090000 +------------------------+
61 | Protected-mode kernel | The bulk of the kernel image.
62 010000 +------------------------+
63 | Boot loader | <- Boot sector entry point 0000:7C00
64 001000 +------------------------+
65 | Reserved for MBR/BIOS |
66 000800 +------------------------+
67 | Typically used by MBR |
68 000600 +------------------------+
70 000000 +------------------------+
73 When using bzImage, the protected-mode kernel was relocated to
74 0x100000 ("high memory"), and the kernel real-mode block (boot sector,
75 setup, and stack/heap) was made relocatable to any address between
76 0x10000 and end of low memory. Unfortunately, in protocols 2.00 and
77 2.01 the command line is still required to live in the 0x9XXXX memory
78 range, and that memory range is still overwritten by the early kernel.
79 The 2.02 protocol resolves that problem.
81 It is desirable to keep the "memory ceiling" -- the highest point in
82 low memory touched by the boot loader -- as low as possible, since
83 some newer BIOSes have begun to allocate some rather large amounts of
84 memory, called the Extended BIOS Data Area, near the top of low
85 memory. The boot loader should use the "INT 12h" BIOS call to verify
86 how much low memory is available.
88 Unfortunately, if INT 12h reports that the amount of memory is too
89 low, there is usually nothing the boot loader can do but to report an
90 error to the user. The boot loader should therefore be designed to
91 take up as little space in low memory as it reasonably can. For
92 zImage or old bzImage kernels, which need data written into the
93 0x90000 segment, the boot loader should make sure not to use memory
94 above the 0x9A000 point; too many BIOSes will break above that point.
97 **** THE REAL-MODE KERNEL HEADER
99 In the following text, and anywhere in the kernel boot sequence, "a
100 sector" refers to 512 bytes. It is independent of the actual sector
101 size of the underlying medium.
103 The first step in loading a Linux kernel should be to load the
104 real-mode code (boot sector and setup code) and then examine the
105 following header at offset 0x01f1. The real-mode code can total up to
106 32K, although the boot loader may choose to load only the first two
107 sectors (1K) and then examine the bootup sector size.
109 The header looks like:
111 Offset Proto Name Meaning
114 01F1/1 ALL(1 setup_sects The size of the setup in sectors
115 01F2/2 ALL root_flags If set, the root is mounted readonly
116 01F4/4 2.04+(2 syssize The size of the 32-bit code in 16-byte paras
117 01F8/2 ALL ram_size DO NOT USE - for bootsect.S use only
118 01FA/2 ALL vid_mode Video mode control
119 01FC/2 ALL root_dev Default root device number
120 01FE/2 ALL boot_flag 0xAA55 magic number
121 0200/2 2.00+ jump Jump instruction
122 0202/4 2.00+ header Magic signature "HdrS"
123 0206/2 2.00+ version Boot protocol version supported
124 0208/4 2.00+ realmode_swtch Boot loader hook (see below)
125 020C/2 2.00+ start_sys The load-low segment (0x1000) (obsolete)
126 020E/2 2.00+ kernel_version Pointer to kernel version string
127 0210/1 2.00+ type_of_loader Boot loader identifier
128 0211/1 2.00+ loadflags Boot protocol option flags
129 0212/2 2.00+ setup_move_size Move to high memory size (used with hooks)
130 0214/4 2.00+ code32_start Boot loader hook (see below)
131 0218/4 2.00+ ramdisk_image initrd load address (set by boot loader)
132 021C/4 2.00+ ramdisk_size initrd size (set by boot loader)
133 0220/4 2.00+ bootsect_kludge DO NOT USE - for bootsect.S use only
134 0224/2 2.01+ heap_end_ptr Free memory after setup end
135 0226/2 N/A pad1 Unused
136 0228/4 2.02+ cmd_line_ptr 32-bit pointer to the kernel command line
137 022C/4 2.03+ initrd_addr_max Highest legal initrd address
138 0230/4 2.05+ kernel_alignment Physical addr alignment required for kernel
139 0234/1 2.05+ relocatable_kernel Whether kernel is relocatable or not
140 0235/3 N/A pad2 Unused
141 0238/4 2.06+ cmdline_size Maximum size of the kernel command line
143 (1) For backwards compatibility, if the setup_sects field contains 0, the
146 (2) For boot protocol prior to 2.04, the upper two bytes of the syssize
147 field are unusable, which means the size of a bzImage kernel
148 cannot be determined.
150 If the "HdrS" (0x53726448) magic number is not found at offset 0x202,
151 the boot protocol version is "old". Loading an old kernel, the
152 following parameters should be assumed:
156 Real-mode kernel must be located at 0x90000.
158 Otherwise, the "version" field contains the protocol version,
159 e.g. protocol version 2.01 will contain 0x0201 in this field. When
160 setting fields in the header, you must make sure only to set fields
161 supported by the protocol version in use.
163 The "kernel_version" field, if set to a nonzero value, contains a
164 pointer to a null-terminated human-readable kernel version number
165 string, less 0x200. This can be used to display the kernel version to
166 the user. This value should be less than (0x200*setup_sects). For
167 example, if this value is set to 0x1c00, the kernel version number
168 string can be found at offset 0x1e00 in the kernel file. This is a
169 valid value if and only if the "setup_sects" field contains the value
172 Most boot loaders will simply load the kernel at its target address
173 directly. Such boot loaders do not need to worry about filling in
174 most of the fields in the header. The following fields should be
178 Please see the section on SPECIAL COMMAND LINE OPTIONS.
181 If your boot loader has an assigned id (see table below), enter
182 0xTV here, where T is an identifier for the boot loader and V is
183 a version number. Otherwise, enter 0xFF here.
185 Assigned boot loader ids:
197 Please contact <hpa@zytor.com> if you need a bootloader ID
200 loadflags, heap_end_ptr:
201 If the protocol version is 2.01 or higher, enter the
202 offset limit of the setup heap into heap_end_ptr and set the
203 0x80 bit (CAN_USE_HEAP) of loadflags. heap_end_ptr appears to
204 be relative to the start of setup (offset 0x0200).
207 When using protocol 2.00 or 2.01, if the real mode
208 kernel is not loaded at 0x90000, it gets moved there later in
209 the loading sequence. Fill in this field if you want
210 additional data (such as the kernel command line) moved in
211 addition to the real-mode kernel itself.
213 ramdisk_image, ramdisk_size:
214 If your boot loader has loaded an initial ramdisk (initrd),
215 set ramdisk_image to the 32-bit pointer to the ramdisk data
216 and the ramdisk_size to the size of the ramdisk data.
218 The initrd should typically be located as high in memory as
219 possible, as it may otherwise get overwritten by the early
220 kernel initialization sequence. However, it must never be
221 located above the address specified in the initrd_addr_max
222 field. The initrd should be at least 4K page aligned.
225 If the protocol version is 2.02 or higher, this is a 32-bit
226 pointer to the kernel command line. The kernel command line
227 can be located anywhere between the end of setup and 0xA0000.
228 Fill in this field even if your boot loader does not support a
229 command line, in which case you can point this to an empty
230 string (or better yet, to the string "auto".) If this field
231 is left at zero, the kernel will assume that your boot loader
232 does not support the 2.02+ protocol.
235 The maximum address that may be occupied by the initrd
236 contents. For boot protocols 2.02 or earlier, this field is
237 not present, and the maximum address is 0x37FFFFFF. (This
238 address is defined as the address of the highest safe byte, so
239 if your ramdisk is exactly 131072 bytes long and this field is
240 0x37FFFFFF, you can start your ramdisk at 0x37FE0000.)
243 The maximum size of the command line without the terminating
244 zero. This means that the command line can contain at most
245 cmdline_size characters. With protocol version 2.05 and
246 earlier, the maximum size was 255.
249 **** THE KERNEL COMMAND LINE
251 The kernel command line has become an important way for the boot
252 loader to communicate with the kernel. Some of its options are also
253 relevant to the boot loader itself, see "special command line options"
256 The kernel command line is a null-terminated string. The maximum
257 length can be retrieved from the field cmdline_size. Before protocol
258 version 2.06, the maximum was 255 characters. A string that is too
259 long will be automatically truncated by the kernel.
261 If the boot protocol version is 2.02 or later, the address of the
262 kernel command line is given by the header field cmd_line_ptr (see
263 above.) This address can be anywhere between the end of the setup
266 If the protocol version is *not* 2.02 or higher, the kernel
267 command line is entered using the following protocol:
269 At offset 0x0020 (word), "cmd_line_magic", enter the magic
272 At offset 0x0022 (word), "cmd_line_offset", enter the offset
273 of the kernel command line (relative to the start of the
276 The kernel command line *must* be within the memory region
277 covered by setup_move_size, so you may need to adjust this
281 **** SAMPLE BOOT CONFIGURATION
283 As a sample configuration, assume the following layout of the real
284 mode segment (this is a typical, and recommended layout):
286 0x0000-0x7FFF Real mode kernel
287 0x8000-0x8FFF Stack and heap
288 0x9000-0x90FF Kernel command line
290 Such a boot loader should enter the following fields in the header:
292 unsigned long base_ptr; /* base address for real-mode segment */
294 if ( setup_sects == 0 ) {
298 if ( protocol >= 0x0200 ) {
299 type_of_loader = <type code>;
300 if ( loading_initrd ) {
301 ramdisk_image = <initrd_address>;
302 ramdisk_size = <initrd_size>;
304 if ( protocol >= 0x0201 ) {
305 heap_end_ptr = 0x9000 - 0x200;
306 loadflags |= 0x80; /* CAN_USE_HEAP */
308 if ( protocol >= 0x0202 ) {
309 cmd_line_ptr = base_ptr + 0x9000;
311 cmd_line_magic = 0xA33F;
312 cmd_line_offset = 0x9000;
313 setup_move_size = 0x9100;
316 /* Very old kernel */
318 cmd_line_magic = 0xA33F;
319 cmd_line_offset = 0x9000;
321 /* A very old kernel MUST have its real-mode code
324 if ( base_ptr != 0x90000 ) {
325 /* Copy the real-mode kernel */
326 memcpy(0x90000, base_ptr, (setup_sects+1)*512);
327 /* Copy the command line */
328 memcpy(0x99000, base_ptr+0x9000, 256);
330 base_ptr = 0x90000; /* Relocated */
333 /* It is recommended to clear memory up to the 32K mark */
334 memset(0x90000 + (setup_sects+1)*512, 0,
335 (64-(setup_sects+1))*512);
339 **** LOADING THE REST OF THE KERNEL
341 The 32-bit (non-real-mode) kernel starts at offset (setup_sects+1)*512
342 in the kernel file (again, if setup_sects == 0 the real value is 4.)
343 It should be loaded at address 0x10000 for Image/zImage kernels and
344 0x100000 for bzImage kernels.
346 The kernel is a bzImage kernel if the protocol >= 2.00 and the 0x01
347 bit (LOAD_HIGH) in the loadflags field is set:
349 is_bzImage = (protocol >= 0x0200) && (loadflags & 0x01);
350 load_address = is_bzImage ? 0x100000 : 0x10000;
352 Note that Image/zImage kernels can be up to 512K in size, and thus use
353 the entire 0x10000-0x90000 range of memory. This means it is pretty
354 much a requirement for these kernels to load the real-mode part at
355 0x90000. bzImage kernels allow much more flexibility.
358 **** SPECIAL COMMAND LINE OPTIONS
360 If the command line provided by the boot loader is entered by the
361 user, the user may expect the following command line options to work.
362 They should normally not be deleted from the kernel command line even
363 though not all of them are actually meaningful to the kernel. Boot
364 loader authors who need additional command line options for the boot
365 loader itself should get them registered in
366 Documentation/kernel-parameters.txt to make sure they will not
367 conflict with actual kernel options now or in the future.
370 <mode> here is either an integer (in C notation, either
371 decimal, octal, or hexadecimal) or one of the strings
372 "normal" (meaning 0xFFFF), "ext" (meaning 0xFFFE) or "ask"
373 (meaning 0xFFFD). This value should be entered into the
374 vid_mode field, as it is used by the kernel before the command
378 <size> is an integer in C notation optionally followed by K, M
379 or G (meaning << 10, << 20 or << 30). This specifies the end
380 of memory to the kernel. This affects the possible placement
381 of an initrd, since an initrd should be placed near end of
382 memory. Note that this is an option to *both* the kernel and
386 An initrd should be loaded. The meaning of <file> is
387 obviously bootloader-dependent, and some boot loaders
388 (e.g. LILO) do not have such a command.
390 In addition, some boot loaders add the following options to the
391 user-specified command line:
394 The boot image which was loaded. Again, the meaning of <file>
395 is obviously bootloader-dependent.
398 The kernel was booted without explicit user intervention.
400 If these options are added by the boot loader, it is highly
401 recommended that they are located *first*, before the user-specified
402 or configuration-specified command line. Otherwise, "init=/bin/sh"
403 gets confused by the "auto" option.
406 **** RUNNING THE KERNEL
408 The kernel is started by jumping to the kernel entry point, which is
409 located at *segment* offset 0x20 from the start of the real mode
410 kernel. This means that if you loaded your real-mode kernel code at
411 0x90000, the kernel entry point is 9020:0000.
413 At entry, ds = es = ss should point to the start of the real-mode
414 kernel code (0x9000 if the code is loaded at 0x90000), sp should be
415 set up properly, normally pointing to the top of the heap, and
416 interrupts should be disabled. Furthermore, to guard against bugs in
417 the kernel, it is recommended that the boot loader sets fs = gs = ds =
420 In our example from above, we would do:
422 /* Note: in the case of the "old" kernel protocol, base_ptr must
423 be == 0x90000 at this point; see the previous sample code */
427 cli(); /* Enter with interrupts disabled! */
429 /* Set up the real-mode kernel stack */
431 _SP = 0x9000; /* Load SP immediately after loading SS! */
433 _DS = _ES = _FS = _GS = seg;
434 jmp_far(seg+0x20, 0); /* Run the kernel */
436 If your boot sector accesses a floppy drive, it is recommended to
437 switch off the floppy motor before running the kernel, since the
438 kernel boot leaves interrupts off and thus the motor will not be
439 switched off, especially if the loaded kernel has the floppy driver as
440 a demand-loaded module!
443 **** ADVANCED BOOT TIME HOOKS
445 If the boot loader runs in a particularly hostile environment (such as
446 LOADLIN, which runs under DOS) it may be impossible to follow the
447 standard memory location requirements. Such a boot loader may use the
448 following hooks that, if set, are invoked by the kernel at the
449 appropriate time. The use of these hooks should probably be
450 considered an absolutely last resort!
452 IMPORTANT: All the hooks are required to preserve %esp, %ebp, %esi and
453 %edi across invocation.
456 A 16-bit real mode far subroutine invoked immediately before
457 entering protected mode. The default routine disables NMI, so
458 your routine should probably do so, too.
461 A 32-bit flat-mode routine *jumped* to immediately after the
462 transition to protected mode, but before the kernel is
463 uncompressed. No segments, except CS, are set up; you should
464 set them up to KERNEL_DS (0x18) yourself.
466 After completing your hook, you should jump to the address
467 that was in this field before your boot loader overwrote it.