aboutsummaryrefslogtreecommitdiffstats
path: root/Documentation/x86
diff options
context:
space:
mode:
authorIngo Molnar <mingo@elte.hu>2008-07-08 03:43:01 -0400
committerIngo Molnar <mingo@elte.hu>2008-07-08 03:43:01 -0400
commit3c1ca43fafea41e38cb2d0c1684119af4c1de547 (patch)
tree122e41a7b9fca26ea25ea9864180f5016274a8c8 /Documentation/x86
parent6924d1ab8b7bbe5ab416713f5701b3316b2df85b (diff)
parent6bcb13b35a2ea39be6c7cc0292b8ad1191b1a748 (diff)
Merge branch 'x86/setup' into x86/devel
Diffstat (limited to 'Documentation/x86')
-rw-r--r--Documentation/x86/i386/IO-APIC.txt119
-rw-r--r--Documentation/x86/i386/boot.txt900
-rw-r--r--Documentation/x86/i386/usb-legacy-support.txt44
-rw-r--r--Documentation/x86/i386/zero-page.txt31
-rw-r--r--Documentation/x86/x86_64/00-INDEX16
-rw-r--r--Documentation/x86/x86_64/boot-options.txt314
-rw-r--r--Documentation/x86/x86_64/cpu-hotplug-spec21
-rw-r--r--Documentation/x86/x86_64/fake-numa-for-cpusets66
-rw-r--r--Documentation/x86/x86_64/kernel-stacks99
-rw-r--r--Documentation/x86/x86_64/machinecheck77
-rw-r--r--Documentation/x86/x86_64/mm.txt28
-rw-r--r--Documentation/x86/x86_64/uefi.txt38
12 files changed, 1753 insertions, 0 deletions
diff --git a/Documentation/x86/i386/IO-APIC.txt b/Documentation/x86/i386/IO-APIC.txt
new file mode 100644
index 000000000000..30b4c714fbe1
--- /dev/null
+++ b/Documentation/x86/i386/IO-APIC.txt
@@ -0,0 +1,119 @@
1Most (all) Intel-MP compliant SMP boards have the so-called 'IO-APIC',
2which is an enhanced interrupt controller. It enables us to route
3hardware interrupts to multiple CPUs, or to CPU groups. Without an
4IO-APIC, interrupts from hardware will be delivered only to the
5CPU which boots the operating system (usually CPU#0).
6
7Linux supports all variants of compliant SMP boards, including ones with
8multiple IO-APICs. Multiple IO-APICs are used in high-end servers to
9distribute IRQ load further.
10
11There are (a few) known breakages in certain older boards, such bugs are
12usually worked around by the kernel. If your MP-compliant SMP board does
13not boot Linux, then consult the linux-smp mailing list archives first.
14
15If your box boots fine with enabled IO-APIC IRQs, then your
16/proc/interrupts will look like this one:
17
18 ---------------------------->
19 hell:~> cat /proc/interrupts
20 CPU0
21 0: 1360293 IO-APIC-edge timer
22 1: 4 IO-APIC-edge keyboard
23 2: 0 XT-PIC cascade
24 13: 1 XT-PIC fpu
25 14: 1448 IO-APIC-edge ide0
26 16: 28232 IO-APIC-level Intel EtherExpress Pro 10/100 Ethernet
27 17: 51304 IO-APIC-level eth0
28 NMI: 0
29 ERR: 0
30 hell:~>
31 <----------------------------
32
33Some interrupts are still listed as 'XT PIC', but this is not a problem;
34none of those IRQ sources is performance-critical.
35
36
37In the unlikely case that your board does not create a working mp-table,
38you can use the pirq= boot parameter to 'hand-construct' IRQ entries. This
39is non-trivial though and cannot be automated. One sample /etc/lilo.conf
40entry:
41
42 append="pirq=15,11,10"
43
44The actual numbers depend on your system, on your PCI cards and on their
45PCI slot position. Usually PCI slots are 'daisy chained' before they are
46connected to the PCI chipset IRQ routing facility (the incoming PIRQ1-4
47lines):
48
49 ,-. ,-. ,-. ,-. ,-.
50 PIRQ4 ----| |-. ,-| |-. ,-| |-. ,-| |--------| |
51 |S| \ / |S| \ / |S| \ / |S| |S|
52 PIRQ3 ----|l|-. `/---|l|-. `/---|l|-. `/---|l|--------|l|
53 |o| \/ |o| \/ |o| \/ |o| |o|
54 PIRQ2 ----|t|-./`----|t|-./`----|t|-./`----|t|--------|t|
55 |1| /\ |2| /\ |3| /\ |4| |5|
56 PIRQ1 ----| |- `----| |- `----| |- `----| |--------| |
57 `-' `-' `-' `-' `-'
58
59Every PCI card emits a PCI IRQ, which can be INTA, INTB, INTC or INTD:
60
61 ,-.
62 INTD--| |
63 |S|
64 INTC--|l|
65 |o|
66 INTB--|t|
67 |x|
68 INTA--| |
69 `-'
70
71These INTA-D PCI IRQs are always 'local to the card', their real meaning
72depends on which slot they are in. If you look at the daisy chaining diagram,
73a card in slot4, issuing INTA IRQ, it will end up as a signal on PIRQ4 of
74the PCI chipset. Most cards issue INTA, this creates optimal distribution
75between the PIRQ lines. (distributing IRQ sources properly is not a
76necessity, PCI IRQs can be shared at will, but it's a good for performance
77to have non shared interrupts). Slot5 should be used for videocards, they
78do not use interrupts normally, thus they are not daisy chained either.
79
80so if you have your SCSI card (IRQ11) in Slot1, Tulip card (IRQ9) in
81Slot2, then you'll have to specify this pirq= line:
82
83 append="pirq=11,9"
84
85the following script tries to figure out such a default pirq= line from
86your PCI configuration:
87
88 echo -n pirq=; echo `scanpci | grep T_L | cut -c56-` | sed 's/ /,/g'
89
90note that this script wont work if you have skipped a few slots or if your
91board does not do default daisy-chaining. (or the IO-APIC has the PIRQ pins
92connected in some strange way). E.g. if in the above case you have your SCSI
93card (IRQ11) in Slot3, and have Slot1 empty:
94
95 append="pirq=0,9,11"
96
97[value '0' is a generic 'placeholder', reserved for empty (or non-IRQ emitting)
98slots.]
99
100Generally, it's always possible to find out the correct pirq= settings, just
101permute all IRQ numbers properly ... it will take some time though. An
102'incorrect' pirq line will cause the booting process to hang, or a device
103won't function properly (e.g. if it's inserted as a module).
104
105If you have 2 PCI buses, then you can use up to 8 pirq values, although such
106boards tend to have a good configuration.
107
108Be prepared that it might happen that you need some strange pirq line:
109
110 append="pirq=0,0,0,0,0,0,9,11"
111
112Use smart trial-and-error techniques to find out the correct pirq line ...
113
114Good luck and mail to linux-smp@vger.kernel.org or
115linux-kernel@vger.kernel.org if you have any problems that are not covered
116by this document.
117
118-- mingo
119
diff --git a/Documentation/x86/i386/boot.txt b/Documentation/x86/i386/boot.txt
new file mode 100644
index 000000000000..147bfe511cdd
--- /dev/null
+++ b/Documentation/x86/i386/boot.txt
@@ -0,0 +1,900 @@
1 THE LINUX/x86 BOOT PROTOCOL
2 ---------------------------
3
4On the x86 platform, the Linux kernel uses a rather complicated boot
5convention. This has evolved partially due to historical aspects, as
6well as the desire in the early days to have the kernel itself be a
7bootable image, the complicated PC memory model and due to changed
8expectations in the PC industry caused by the effective demise of
9real-mode DOS as a mainstream operating system.
10
11Currently, the following versions of the Linux/x86 boot protocol exist.
12
13Old kernels: zImage/Image support only. Some very early kernels
14 may not even support a command line.
15
16Protocol 2.00: (Kernel 1.3.73) Added bzImage and initrd support, as
17 well as a formalized way to communicate between the
18 boot loader and the kernel. setup.S made relocatable,
19 although the traditional setup area still assumed
20 writable.
21
22Protocol 2.01: (Kernel 1.3.76) Added a heap overrun warning.
23
24Protocol 2.02: (Kernel 2.4.0-test3-pre3) New command line protocol.
25 Lower the conventional memory ceiling. No overwrite
26 of the traditional setup area, thus making booting
27 safe for systems which use the EBDA from SMM or 32-bit
28 BIOS entry points. zImage deprecated but still
29 supported.
30
31Protocol 2.03: (Kernel 2.4.18-pre1) Explicitly makes the highest possible
32 initrd address available to the bootloader.
33
34Protocol 2.04: (Kernel 2.6.14) Extend the syssize field to four bytes.
35
36Protocol 2.05: (Kernel 2.6.20) Make protected mode kernel relocatable.
37 Introduce relocatable_kernel and kernel_alignment fields.
38
39Protocol 2.06: (Kernel 2.6.22) Added a field that contains the size of
40 the boot command line.
41
42Protocol 2.07: (Kernel 2.6.24) Added paravirtualised boot protocol.
43 Introduced hardware_subarch and hardware_subarch_data
44 and KEEP_SEGMENTS flag in load_flags.
45
46Protocol 2.08: (Kernel 2.6.26) Added crc32 checksum and ELF format
47 payload. Introduced payload_offset and payload length
48 fields to aid in locating the payload.
49
50Protocol 2.09: (Kernel 2.6.26) Added a field of 64-bit physical
51 pointer to single linked list of struct setup_data.
52
53**** MEMORY LAYOUT
54
55The traditional memory map for the kernel loader, used for Image or
56zImage kernels, typically looks like:
57
58 | |
590A0000 +------------------------+
60 | Reserved for BIOS | Do not use. Reserved for BIOS EBDA.
6109A000 +------------------------+
62 | Command line |
63 | Stack/heap | For use by the kernel real-mode code.
64098000 +------------------------+
65 | Kernel setup | The kernel real-mode code.
66090200 +------------------------+
67 | Kernel boot sector | The kernel legacy boot sector.
68090000 +------------------------+
69 | Protected-mode kernel | The bulk of the kernel image.
70010000 +------------------------+
71 | Boot loader | <- Boot sector entry point 0000:7C00
72001000 +------------------------+
73 | Reserved for MBR/BIOS |
74000800 +------------------------+
75 | Typically used by MBR |
76000600 +------------------------+
77 | BIOS use only |
78000000 +------------------------+
79
80
81When using bzImage, the protected-mode kernel was relocated to
820x100000 ("high memory"), and the kernel real-mode block (boot sector,
83setup, and stack/heap) was made relocatable to any address between
840x10000 and end of low memory. Unfortunately, in protocols 2.00 and
852.01 the 0x90000+ memory range is still used internally by the kernel;
86the 2.02 protocol resolves that problem.
87
88It is desirable to keep the "memory ceiling" -- the highest point in
89low memory touched by the boot loader -- as low as possible, since
90some newer BIOSes have begun to allocate some rather large amounts of
91memory, called the Extended BIOS Data Area, near the top of low
92memory. The boot loader should use the "INT 12h" BIOS call to verify
93how much low memory is available.
94
95Unfortunately, if INT 12h reports that the amount of memory is too
96low, there is usually nothing the boot loader can do but to report an
97error to the user. The boot loader should therefore be designed to
98take up as little space in low memory as it reasonably can. For
99zImage or old bzImage kernels, which need data written into the
1000x90000 segment, the boot loader should make sure not to use memory
101above the 0x9A000 point; too many BIOSes will break above that point.
102
103For a modern bzImage kernel with boot protocol version >= 2.02, a
104memory layout like the following is suggested:
105
106 ~ ~
107 | Protected-mode kernel |
108100000 +------------------------+
109 | I/O memory hole |
1100A0000 +------------------------+
111 | Reserved for BIOS | Leave as much as possible unused
112 ~ ~
113 | Command line | (Can also be below the X+10000 mark)
114X+10000 +------------------------+
115 | Stack/heap | For use by the kernel real-mode code.
116X+08000 +------------------------+
117 | Kernel setup | The kernel real-mode code.
118 | Kernel boot sector | The kernel legacy boot sector.
119X +------------------------+
120 | Boot loader | <- Boot sector entry point 0000:7C00
121001000 +------------------------+
122 | Reserved for MBR/BIOS |
123000800 +------------------------+
124 | Typically used by MBR |
125000600 +------------------------+
126 | BIOS use only |
127000000 +------------------------+
128
129... where the address X is as low as the design of the boot loader
130permits.
131
132
133**** THE REAL-MODE KERNEL HEADER
134
135In the following text, and anywhere in the kernel boot sequence, "a
136sector" refers to 512 bytes. It is independent of the actual sector
137size of the underlying medium.
138
139The first step in loading a Linux kernel should be to load the
140real-mode code (boot sector and setup code) and then examine the
141following header at offset 0x01f1. The real-mode code can total up to
14232K, although the boot loader may choose to load only the first two
143sectors (1K) and then examine the bootup sector size.
144
145The header looks like:
146
147Offset Proto Name Meaning
148/Size
149
15001F1/1 ALL(1 setup_sects The size of the setup in sectors
15101F2/2 ALL root_flags If set, the root is mounted readonly
15201F4/4 2.04+(2 syssize The size of the 32-bit code in 16-byte paras
15301F8/2 ALL ram_size DO NOT USE - for bootsect.S use only
15401FA/2 ALL vid_mode Video mode control
15501FC/2 ALL root_dev Default root device number
15601FE/2 ALL boot_flag 0xAA55 magic number
1570200/2 2.00+ jump Jump instruction
1580202/4 2.00+ header Magic signature "HdrS"
1590206/2 2.00+ version Boot protocol version supported
1600208/4 2.00+ realmode_swtch Boot loader hook (see below)
161020C/2 2.00+ start_sys The load-low segment (0x1000) (obsolete)
162020E/2 2.00+ kernel_version Pointer to kernel version string
1630210/1 2.00+ type_of_loader Boot loader identifier
1640211/1 2.00+ loadflags Boot protocol option flags
1650212/2 2.00+ setup_move_size Move to high memory size (used with hooks)
1660214/4 2.00+ code32_start Boot loader hook (see below)
1670218/4 2.00+ ramdisk_image initrd load address (set by boot loader)
168021C/4 2.00+ ramdisk_size initrd size (set by boot loader)
1690220/4 2.00+ bootsect_kludge DO NOT USE - for bootsect.S use only
1700224/2 2.01+ heap_end_ptr Free memory after setup end
1710226/2 N/A pad1 Unused
1720228/4 2.02+ cmd_line_ptr 32-bit pointer to the kernel command line
173022C/4 2.03+ initrd_addr_max Highest legal initrd address
1740230/4 2.05+ kernel_alignment Physical addr alignment required for kernel
1750234/1 2.05+ relocatable_kernel Whether kernel is relocatable or not
1760235/3 N/A pad2 Unused
1770238/4 2.06+ cmdline_size Maximum size of the kernel command line
178023C/4 2.07+ hardware_subarch Hardware subarchitecture
1790240/8 2.07+ hardware_subarch_data Subarchitecture-specific data
1800248/4 2.08+ payload_offset Offset of kernel payload
181024C/4 2.08+ payload_length Length of kernel payload
1820250/8 2.09+ setup_data 64-bit physical pointer to linked list
183 of struct setup_data
184
185(1) For backwards compatibility, if the setup_sects field contains 0, the
186 real value is 4.
187
188(2) For boot protocol prior to 2.04, the upper two bytes of the syssize
189 field are unusable, which means the size of a bzImage kernel
190 cannot be determined.
191
192If the "HdrS" (0x53726448) magic number is not found at offset 0x202,
193the boot protocol version is "old". Loading an old kernel, the
194following parameters should be assumed:
195
196 Image type = zImage
197 initrd not supported
198 Real-mode kernel must be located at 0x90000.
199
200Otherwise, the "version" field contains the protocol version,
201e.g. protocol version 2.01 will contain 0x0201 in this field. When
202setting fields in the header, you must make sure only to set fields
203supported by the protocol version in use.
204
205
206**** DETAILS OF HEADER FIELDS
207
208For each field, some are information from the kernel to the bootloader
209("read"), some are expected to be filled out by the bootloader
210("write"), and some are expected to be read and modified by the
211bootloader ("modify").
212
213All general purpose boot loaders should write the fields marked
214(obligatory). Boot loaders who want to load the kernel at a
215nonstandard address should fill in the fields marked (reloc); other
216boot loaders can ignore those fields.
217
218The byte order of all fields is littleendian (this is x86, after all.)
219
220Field name: setup_sects
221Type: read
222Offset/size: 0x1f1/1
223Protocol: ALL
224
225 The size of the setup code in 512-byte sectors. If this field is
226 0, the real value is 4. The real-mode code consists of the boot
227 sector (always one 512-byte sector) plus the setup code.
228
229Field name: root_flags
230Type: modify (optional)
231Offset/size: 0x1f2/2
232Protocol: ALL
233
234 If this field is nonzero, the root defaults to readonly. The use of
235 this field is deprecated; use the "ro" or "rw" options on the
236 command line instead.
237
238Field name: syssize
239Type: read
240Offset/size: 0x1f4/4 (protocol 2.04+) 0x1f4/2 (protocol ALL)
241Protocol: 2.04+
242
243 The size of the protected-mode code in units of 16-byte paragraphs.
244 For protocol versions older than 2.04 this field is only two bytes
245 wide, and therefore cannot be trusted for the size of a kernel if
246 the LOAD_HIGH flag is set.
247
248Field name: ram_size
249Type: kernel internal
250Offset/size: 0x1f8/2
251Protocol: ALL
252
253 This field is obsolete.
254
255Field name: vid_mode
256Type: modify (obligatory)
257Offset/size: 0x1fa/2
258
259 Please see the section on SPECIAL COMMAND LINE OPTIONS.
260
261Field name: root_dev
262Type: modify (optional)
263Offset/size: 0x1fc/2
264Protocol: ALL
265
266 The default root device device number. The use of this field is
267 deprecated, use the "root=" option on the command line instead.
268
269Field name: boot_flag
270Type: read
271Offset/size: 0x1fe/2
272Protocol: ALL
273
274 Contains 0xAA55. This is the closest thing old Linux kernels have
275 to a magic number.
276
277Field name: jump
278Type: read
279Offset/size: 0x200/2
280Protocol: 2.00+
281
282 Contains an x86 jump instruction, 0xEB followed by a signed offset
283 relative to byte 0x202. This can be used to determine the size of
284 the header.
285
286Field name: header
287Type: read
288Offset/size: 0x202/4
289Protocol: 2.00+
290
291 Contains the magic number "HdrS" (0x53726448).
292
293Field name: version
294Type: read
295Offset/size: 0x206/2
296Protocol: 2.00+
297
298 Contains the boot protocol version, in (major << 8)+minor format,
299 e.g. 0x0204 for version 2.04, and 0x0a11 for a hypothetical version
300 10.17.
301
302Field name: readmode_swtch
303Type: modify (optional)
304Offset/size: 0x208/4
305Protocol: 2.00+
306
307 Boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
308
309Field name: start_sys
310Type: read
311Offset/size: 0x20c/4
312Protocol: 2.00+
313
314 The load low segment (0x1000). Obsolete.
315
316Field name: kernel_version
317Type: read
318Offset/size: 0x20e/2
319Protocol: 2.00+
320
321 If set to a nonzero value, contains a pointer to a NUL-terminated
322 human-readable kernel version number string, less 0x200. This can
323 be used to display the kernel version to the user. This value
324 should be less than (0x200*setup_sects).
325
326 For example, if this value is set to 0x1c00, the kernel version
327 number string can be found at offset 0x1e00 in the kernel file.
328 This is a valid value if and only if the "setup_sects" field
329 contains the value 15 or higher, as:
330
331 0x1c00 < 15*0x200 (= 0x1e00) but
332 0x1c00 >= 14*0x200 (= 0x1c00)
333
334 0x1c00 >> 9 = 14, so the minimum value for setup_secs is 15.
335
336Field name: type_of_loader
337Type: write (obligatory)
338Offset/size: 0x210/1
339Protocol: 2.00+
340
341 If your boot loader has an assigned id (see table below), enter
342 0xTV here, where T is an identifier for the boot loader and V is
343 a version number. Otherwise, enter 0xFF here.
344
345 Assigned boot loader ids:
346 0 LILO (0x00 reserved for pre-2.00 bootloader)
347 1 Loadlin
348 2 bootsect-loader (0x20, all other values reserved)
349 3 SYSLINUX
350 4 EtherBoot
351 5 ELILO
352 7 GRuB
353 8 U-BOOT
354 9 Xen
355 A Gujin
356 B Qemu
357
358 Please contact <hpa@zytor.com> if you need a bootloader ID
359 value assigned.
360
361Field name: loadflags
362Type: modify (obligatory)
363Offset/size: 0x211/1
364Protocol: 2.00+
365
366 This field is a bitmask.
367
368 Bit 0 (read): LOADED_HIGH
369 - If 0, the protected-mode code is loaded at 0x10000.
370 - If 1, the protected-mode code is loaded at 0x100000.
371
372 Bit 5 (write): QUIET_FLAG
373 - If 0, print early messages.
374 - If 1, suppress early messages.
375 This requests to the kernel (decompressor and early
376 kernel) to not write early messages that require
377 accessing the display hardware directly.
378
379 Bit 6 (write): KEEP_SEGMENTS
380 Protocol: 2.07+
381 - If 0, reload the segment registers in the 32bit entry point.
382 - If 1, do not reload the segment registers in the 32bit entry point.
383 Assume that %cs %ds %ss %es are all set to flat segments with
384 a base of 0 (or the equivalent for their environment).
385
386 Bit 7 (write): CAN_USE_HEAP
387 Set this bit to 1 to indicate that the value entered in the
388 heap_end_ptr is valid. If this field is clear, some setup code
389 functionality will be disabled.
390
391Field name: setup_move_size
392Type: modify (obligatory)
393Offset/size: 0x212/2
394Protocol: 2.00-2.01
395
396 When using protocol 2.00 or 2.01, if the real mode kernel is not
397 loaded at 0x90000, it gets moved there later in the loading
398 sequence. Fill in this field if you want additional data (such as
399 the kernel command line) moved in addition to the real-mode kernel
400 itself.
401
402 The unit is bytes starting with the beginning of the boot sector.
403
404 This field is can be ignored when the protocol is 2.02 or higher, or
405 if the real-mode code is loaded at 0x90000.
406
407Field name: code32_start
408Type: modify (optional, reloc)
409Offset/size: 0x214/4
410Protocol: 2.00+
411
412 The address to jump to in protected mode. This defaults to the load
413 address of the kernel, and can be used by the boot loader to
414 determine the proper load address.
415
416 This field can be modified for two purposes:
417
418 1. as a boot loader hook (see ADVANCED BOOT LOADER HOOKS below.)
419
420 2. if a bootloader which does not install a hook loads a
421 relocatable kernel at a nonstandard address it will have to modify
422 this field to point to the load address.
423
424Field name: ramdisk_image
425Type: write (obligatory)
426Offset/size: 0x218/4
427Protocol: 2.00+
428
429 The 32-bit linear address of the initial ramdisk or ramfs. Leave at
430 zero if there is no initial ramdisk/ramfs.
431
432Field name: ramdisk_size
433Type: write (obligatory)
434Offset/size: 0x21c/4
435Protocol: 2.00+
436
437 Size of the initial ramdisk or ramfs. Leave at zero if there is no
438 initial ramdisk/ramfs.
439
440Field name: bootsect_kludge
441Type: kernel internal
442Offset/size: 0x220/4
443Protocol: 2.00+
444
445 This field is obsolete.
446
447Field name: heap_end_ptr
448Type: write (obligatory)
449Offset/size: 0x224/2
450Protocol: 2.01+
451
452 Set this field to the offset (from the beginning of the real-mode
453 code) of the end of the setup stack/heap, minus 0x0200.
454
455Field name: cmd_line_ptr
456Type: write (obligatory)
457Offset/size: 0x228/4
458Protocol: 2.02+
459
460 Set this field to the linear address of the kernel command line.
461 The kernel command line can be located anywhere between the end of
462 the setup heap and 0xA0000; it does not have to be located in the
463 same 64K segment as the real-mode code itself.
464
465 Fill in this field even if your boot loader does not support a
466 command line, in which case you can point this to an empty string
467 (or better yet, to the string "auto".) If this field is left at
468 zero, the kernel will assume that your boot loader does not support
469 the 2.02+ protocol.
470
471Field name: initrd_addr_max
472Type: read
473Offset/size: 0x22c/4
474Protocol: 2.03+
475
476 The maximum address that may be occupied by the initial
477 ramdisk/ramfs contents. For boot protocols 2.02 or earlier, this
478 field is not present, and the maximum address is 0x37FFFFFF. (This
479 address is defined as the address of the highest safe byte, so if
480 your ramdisk is exactly 131072 bytes long and this field is
481 0x37FFFFFF, you can start your ramdisk at 0x37FE0000.)
482
483Field name: kernel_alignment
484Type: read (reloc)
485Offset/size: 0x230/4
486Protocol: 2.05+
487
488 Alignment unit required by the kernel (if relocatable_kernel is true.)
489
490Field name: relocatable_kernel
491Type: read (reloc)
492Offset/size: 0x234/1
493Protocol: 2.05+
494
495 If this field is nonzero, the protected-mode part of the kernel can
496 be loaded at any address that satisfies the kernel_alignment field.
497 After loading, the boot loader must set the code32_start field to
498 point to the loaded code, or to a boot loader hook.
499
500Field name: cmdline_size
501Type: read
502Offset/size: 0x238/4
503Protocol: 2.06+
504
505 The maximum size of the command line without the terminating
506 zero. This means that the command line can contain at most
507 cmdline_size characters. With protocol version 2.05 and earlier, the
508 maximum size was 255.
509
510Field name: hardware_subarch
511Type: write (optional, defaults to x86/PC)
512Offset/size: 0x23c/4
513Protocol: 2.07+
514
515 In a paravirtualized environment the hardware low level architectural
516 pieces such as interrupt handling, page table handling, and
517 accessing process control registers needs to be done differently.
518
519 This field allows the bootloader to inform the kernel we are in one
520 one of those environments.
521
522 0x00000000 The default x86/PC environment
523 0x00000001 lguest
524 0x00000002 Xen
525
526Field name: hardware_subarch_data
527Type: write (subarch-dependent)
528Offset/size: 0x240/8
529Protocol: 2.07+
530
531 A pointer to data that is specific to hardware subarch
532 This field is currently unused for the default x86/PC environment,
533 do not modify.
534
535Field name: payload_offset
536Type: read
537Offset/size: 0x248/4
538Protocol: 2.08+
539
540 If non-zero then this field contains the offset from the end of the
541 real-mode code to the payload.
542
543 The payload may be compressed. The format of both the compressed and
544 uncompressed data should be determined using the standard magic
545 numbers. Currently only gzip compressed ELF is used.
546
547Field name: payload_length
548Type: read
549Offset/size: 0x24c/4
550Protocol: 2.08+
551
552 The length of the payload.
553
554Field name: setup_data
555Type: write (special)
556Offset/size: 0x250/8
557Protocol: 2.09+
558
559 The 64-bit physical pointer to NULL terminated single linked list of
560 struct setup_data. This is used to define a more extensible boot
561 parameters passing mechanism. The definition of struct setup_data is
562 as follow:
563
564 struct setup_data {
565 u64 next;
566 u32 type;
567 u32 len;
568 u8 data[0];
569 };
570
571 Where, the next is a 64-bit physical pointer to the next node of
572 linked list, the next field of the last node is 0; the type is used
573 to identify the contents of data; the len is the length of data
574 field; the data holds the real payload.
575
576 This list may be modified at a number of points during the bootup
577 process. Therefore, when modifying this list one should always make
578 sure to consider the case where the linked list already contains
579 entries.
580
581
582**** THE IMAGE CHECKSUM
583
584From boot protocol version 2.08 onwards the CRC-32 is calculated over
585the entire file using the characteristic polynomial 0x04C11DB7 and an
586initial remainder of 0xffffffff. The checksum is appended to the
587file; therefore the CRC of the file up to the limit specified in the
588syssize field of the header is always 0.
589
590
591**** THE KERNEL COMMAND LINE
592
593The kernel command line has become an important way for the boot
594loader to communicate with the kernel. Some of its options are also
595relevant to the boot loader itself, see "special command line options"
596below.
597
598The kernel command line is a null-terminated string. The maximum
599length can be retrieved from the field cmdline_size. Before protocol
600version 2.06, the maximum was 255 characters. A string that is too
601long will be automatically truncated by the kernel.
602
603If the boot protocol version is 2.02 or later, the address of the
604kernel command line is given by the header field cmd_line_ptr (see
605above.) This address can be anywhere between the end of the setup
606heap and 0xA0000.
607
608If the protocol version is *not* 2.02 or higher, the kernel
609command line is entered using the following protocol:
610
611 At offset 0x0020 (word), "cmd_line_magic", enter the magic
612 number 0xA33F.
613
614 At offset 0x0022 (word), "cmd_line_offset", enter the offset
615 of the kernel command line (relative to the start of the
616 real-mode kernel).
617
618 The kernel command line *must* be within the memory region
619 covered by setup_move_size, so you may need to adjust this
620 field.
621
622
623**** MEMORY LAYOUT OF THE REAL-MODE CODE
624
625The real-mode code requires a stack/heap to be set up, as well as
626memory allocated for the kernel command line. This needs to be done
627in the real-mode accessible memory in bottom megabyte.
628
629It should be noted that modern machines often have a sizable Extended
630BIOS Data Area (EBDA). As a result, it is advisable to use as little
631of the low megabyte as possible.
632
633Unfortunately, under the following circumstances the 0x90000 memory
634segment has to be used:
635
636 - When loading a zImage kernel ((loadflags & 0x01) == 0).
637 - When loading a 2.01 or earlier boot protocol kernel.
638
639 -> For the 2.00 and 2.01 boot protocols, the real-mode code
640 can be loaded at another address, but it is internally
641 relocated to 0x90000. For the "old" protocol, the
642 real-mode code must be loaded at 0x90000.
643
644When loading at 0x90000, avoid using memory above 0x9a000.
645
646For boot protocol 2.02 or higher, the command line does not have to be
647located in the same 64K segment as the real-mode setup code; it is
648thus permitted to give the stack/heap the full 64K segment and locate
649the command line above it.
650
651The kernel command line should not be located below the real-mode
652code, nor should it be located in high memory.
653
654
655**** SAMPLE BOOT CONFIGURATION
656
657As a sample configuration, assume the following layout of the real
658mode segment:
659
660 When loading below 0x90000, use the entire segment:
661
662 0x0000-0x7fff Real mode kernel
663 0x8000-0xdfff Stack and heap
664 0xe000-0xffff Kernel command line
665
666 When loading at 0x90000 OR the protocol version is 2.01 or earlier:
667
668 0x0000-0x7fff Real mode kernel
669 0x8000-0x97ff Stack and heap
670 0x9800-0x9fff Kernel command line
671
672Such a boot loader should enter the following fields in the header:
673
674 unsigned long base_ptr; /* base address for real-mode segment */
675
676 if ( setup_sects == 0 ) {
677 setup_sects = 4;
678 }
679
680 if ( protocol >= 0x0200 ) {
681 type_of_loader = <type code>;
682 if ( loading_initrd ) {
683 ramdisk_image = <initrd_address>;
684 ramdisk_size = <initrd_size>;
685 }
686
687 if ( protocol >= 0x0202 && loadflags & 0x01 )
688 heap_end = 0xe000;
689 else
690 heap_end = 0x9800;
691
692 if ( protocol >= 0x0201 ) {
693 heap_end_ptr = heap_end - 0x200;
694 loadflags |= 0x80; /* CAN_USE_HEAP */
695 }
696
697 if ( protocol >= 0x0202 ) {
698 cmd_line_ptr = base_ptr + heap_end;
699 strcpy(cmd_line_ptr, cmdline);
700 } else {
701 cmd_line_magic = 0xA33F;
702 cmd_line_offset = heap_end;
703 setup_move_size = heap_end + strlen(cmdline)+1;
704 strcpy(base_ptr+cmd_line_offset, cmdline);
705 }
706 } else {
707 /* Very old kernel */
708
709 heap_end = 0x9800;
710
711 cmd_line_magic = 0xA33F;
712 cmd_line_offset = heap_end;
713
714 /* A very old kernel MUST have its real-mode code
715 loaded at 0x90000 */
716
717 if ( base_ptr != 0x90000 ) {
718 /* Copy the real-mode kernel */
719 memcpy(0x90000, base_ptr, (setup_sects+1)*512);
720 base_ptr = 0x90000; /* Relocated */
721 }
722
723 strcpy(0x90000+cmd_line_offset, cmdline);
724
725 /* It is recommended to clear memory up to the 32K mark */
726 memset(0x90000 + (setup_sects+1)*512, 0,
727 (64-(setup_sects+1))*512);
728 }
729
730
731**** LOADING THE REST OF THE KERNEL
732
733The 32-bit (non-real-mode) kernel starts at offset (setup_sects+1)*512
734in the kernel file (again, if setup_sects == 0 the real value is 4.)
735It should be loaded at address 0x10000 for Image/zImage kernels and
7360x100000 for bzImage kernels.
737
738The kernel is a bzImage kernel if the protocol >= 2.00 and the 0x01
739bit (LOAD_HIGH) in the loadflags field is set:
740
741 is_bzImage = (protocol >= 0x0200) && (loadflags & 0x01);
742 load_address = is_bzImage ? 0x100000 : 0x10000;
743
744Note that Image/zImage kernels can be up to 512K in size, and thus use
745the entire 0x10000-0x90000 range of memory. This means it is pretty
746much a requirement for these kernels to load the real-mode part at
7470x90000. bzImage kernels allow much more flexibility.
748
749
750**** SPECIAL COMMAND LINE OPTIONS
751
752If the command line provided by the boot loader is entered by the
753user, the user may expect the following command line options to work.
754They should normally not be deleted from the kernel command line even
755though not all of them are actually meaningful to the kernel. Boot
756loader authors who need additional command line options for the boot
757loader itself should get them registered in
758Documentation/kernel-parameters.txt to make sure they will not
759conflict with actual kernel options now or in the future.
760
761 vga=<mode>
762 <mode> here is either an integer (in C notation, either
763 decimal, octal, or hexadecimal) or one of the strings
764 "normal" (meaning 0xFFFF), "ext" (meaning 0xFFFE) or "ask"
765 (meaning 0xFFFD). This value should be entered into the
766 vid_mode field, as it is used by the kernel before the command
767 line is parsed.
768
769 mem=<size>
770 <size> is an integer in C notation optionally followed by
771 (case insensitive) K, M, G, T, P or E (meaning << 10, << 20,
772 << 30, << 40, << 50 or << 60). This specifies the end of
773 memory to the kernel. This affects the possible placement of
774 an initrd, since an initrd should be placed near end of
775 memory. Note that this is an option to *both* the kernel and
776 the bootloader!
777
778 initrd=<file>
779 An initrd should be loaded. The meaning of <file> is
780 obviously bootloader-dependent, and some boot loaders
781 (e.g. LILO) do not have such a command.
782
783In addition, some boot loaders add the following options to the
784user-specified command line:
785
786 BOOT_IMAGE=<file>
787 The boot image which was loaded. Again, the meaning of <file>
788 is obviously bootloader-dependent.
789
790 auto
791 The kernel was booted without explicit user intervention.
792
793If these options are added by the boot loader, it is highly
794recommended that they are located *first*, before the user-specified
795or configuration-specified command line. Otherwise, "init=/bin/sh"
796gets confused by the "auto" option.
797
798
799**** RUNNING THE KERNEL
800
801The kernel is started by jumping to the kernel entry point, which is
802located at *segment* offset 0x20 from the start of the real mode
803kernel. This means that if you loaded your real-mode kernel code at
8040x90000, the kernel entry point is 9020:0000.
805
806At entry, ds = es = ss should point to the start of the real-mode
807kernel code (0x9000 if the code is loaded at 0x90000), sp should be
808set up properly, normally pointing to the top of the heap, and
809interrupts should be disabled. Furthermore, to guard against bugs in
810the kernel, it is recommended that the boot loader sets fs = gs = ds =
811es = ss.
812
813In our example from above, we would do:
814
815 /* Note: in the case of the "old" kernel protocol, base_ptr must
816 be == 0x90000 at this point; see the previous sample code */
817
818 seg = base_ptr >> 4;
819
820 cli(); /* Enter with interrupts disabled! */
821
822 /* Set up the real-mode kernel stack */
823 _SS = seg;
824 _SP = heap_end;
825
826 _DS = _ES = _FS = _GS = seg;
827 jmp_far(seg+0x20, 0); /* Run the kernel */
828
829If your boot sector accesses a floppy drive, it is recommended to
830switch off the floppy motor before running the kernel, since the
831kernel boot leaves interrupts off and thus the motor will not be
832switched off, especially if the loaded kernel has the floppy driver as
833a demand-loaded module!
834
835
836**** ADVANCED BOOT LOADER HOOKS
837
838If the boot loader runs in a particularly hostile environment (such as
839LOADLIN, which runs under DOS) it may be impossible to follow the
840standard memory location requirements. Such a boot loader may use the
841following hooks that, if set, are invoked by the kernel at the
842appropriate time. The use of these hooks should probably be
843considered an absolutely last resort!
844
845IMPORTANT: All the hooks are required to preserve %esp, %ebp, %esi and
846%edi across invocation.
847
848 realmode_swtch:
849 A 16-bit real mode far subroutine invoked immediately before
850 entering protected mode. The default routine disables NMI, so
851 your routine should probably do so, too.
852
853 code32_start:
854 A 32-bit flat-mode routine *jumped* to immediately after the
855 transition to protected mode, but before the kernel is
856 uncompressed. No segments, except CS, are guaranteed to be
857 set up (current kernels do, but older ones do not); you should
858 set them up to BOOT_DS (0x18) yourself.
859
860 After completing your hook, you should jump to the address
861 that was in this field before your boot loader overwrote it
862 (relocated, if appropriate.)
863
864
865**** 32-bit BOOT PROTOCOL
866
867For machine with some new BIOS other than legacy BIOS, such as EFI,
868LinuxBIOS, etc, and kexec, the 16-bit real mode setup code in kernel
869based on legacy BIOS can not be used, so a 32-bit boot protocol needs
870to be defined.
871
872In 32-bit boot protocol, the first step in loading a Linux kernel
873should be to setup the boot parameters (struct boot_params,
874traditionally known as "zero page"). The memory for struct boot_params
875should be allocated and initialized to all zero. Then the setup header
876from offset 0x01f1 of kernel image on should be loaded into struct
877boot_params and examined. The end of setup header can be calculated as
878follow:
879
880 0x0202 + byte value at offset 0x0201
881
882In addition to read/modify/write the setup header of the struct
883boot_params as that of 16-bit boot protocol, the boot loader should
884also fill the additional fields of the struct boot_params as that
885described in zero-page.txt.
886
887After setupping the struct boot_params, the boot loader can load the
88832/64-bit kernel in the same way as that of 16-bit boot protocol.
889
890In 32-bit boot protocol, the kernel is started by jumping to the
89132-bit kernel entry point, which is the start address of loaded
89232/64-bit kernel.
893
894At entry, the CPU must be in 32-bit protected mode with paging
895disabled; a GDT must be loaded with the descriptors for selectors
896__BOOT_CS(0x10) and __BOOT_DS(0x18); both descriptors must be 4G flat
897segment; __BOOS_CS must have execute/read permission, and __BOOT_DS
898must have read/write permission; CS must be __BOOT_CS and DS, ES, SS
899must be __BOOT_DS; interrupt must be disabled; %esi must hold the base
900address of the struct boot_params; %ebp, %edi and %ebx must be zero.
diff --git a/Documentation/x86/i386/usb-legacy-support.txt b/Documentation/x86/i386/usb-legacy-support.txt
new file mode 100644
index 000000000000..1894cdfc69d9
--- /dev/null
+++ b/Documentation/x86/i386/usb-legacy-support.txt
@@ -0,0 +1,44 @@
1USB Legacy support
2~~~~~~~~~~~~~~~~~~
3
4Vojtech Pavlik <vojtech@suse.cz>, January 2004
5
6
7Also known as "USB Keyboard" or "USB Mouse support" in the BIOS Setup is a
8feature that allows one to use the USB mouse and keyboard as if they were
9their classic PS/2 counterparts. This means one can use an USB keyboard to
10type in LILO for example.
11
12It has several drawbacks, though:
13
141) On some machines, the emulated PS/2 mouse takes over even when no USB
15 mouse is present and a real PS/2 mouse is present. In that case the extra
16 features (wheel, extra buttons, touchpad mode) of the real PS/2 mouse may
17 not be available.
18
192) If CONFIG_HIGHMEM64G is enabled, the PS/2 mouse emulation can cause
20 system crashes, because the SMM BIOS is not expecting to be in PAE mode.
21 The Intel E7505 is a typical machine where this happens.
22
233) If AMD64 64-bit mode is enabled, again system crashes often happen,
24 because the SMM BIOS isn't expecting the CPU to be in 64-bit mode. The
25 BIOS manufacturers only test with Windows, and Windows doesn't do 64-bit
26 yet.
27
28Solutions:
29
30Problem 1) can be solved by loading the USB drivers prior to loading the
31PS/2 mouse driver. Since the PS/2 mouse driver is in 2.6 compiled into
32the kernel unconditionally, this means the USB drivers need to be
33compiled-in, too.
34
35Problem 2) can currently only be solved by either disabling HIGHMEM64G
36in the kernel config or USB Legacy support in the BIOS. A BIOS update
37could help, but so far no such update exists.
38
39Problem 3) is usually fixed by a BIOS update. Check the board
40manufacturers web site. If an update is not available, disable USB
41Legacy support in the BIOS. If this alone doesn't help, try also adding
42idle=poll on the kernel command line. The BIOS may be entering the SMM
43on the HLT instruction as well.
44
diff --git a/Documentation/x86/i386/zero-page.txt b/Documentation/x86/i386/zero-page.txt
new file mode 100644
index 000000000000..169ad423a3d1
--- /dev/null
+++ b/Documentation/x86/i386/zero-page.txt
@@ -0,0 +1,31 @@
1The additional fields in struct boot_params as a part of 32-bit boot
2protocol of kernel. These should be filled by bootloader or 16-bit
3real-mode setup code of the kernel. References/settings to it mainly
4are in:
5
6 include/asm-x86/bootparam.h
7
8
9Offset Proto Name Meaning
10/Size
11
12000/040 ALL screen_info Text mode or frame buffer information
13 (struct screen_info)
14040/014 ALL apm_bios_info APM BIOS information (struct apm_bios_info)
15060/010 ALL ist_info Intel SpeedStep (IST) BIOS support information
16 (struct ist_info)
17080/010 ALL hd0_info hd0 disk parameter, OBSOLETE!!
18090/010 ALL hd1_info hd1 disk parameter, OBSOLETE!!
190A0/010 ALL sys_desc_table System description table (struct sys_desc_table)
20140/080 ALL edid_info Video mode setup (struct edid_info)
211C0/020 ALL efi_info EFI 32 information (struct efi_info)
221E0/004 ALL alk_mem_k Alternative mem check, in KB
231E4/004 ALL scratch Scratch field for the kernel setup code
241E8/001 ALL e820_entries Number of entries in e820_map (below)
251E9/001 ALL eddbuf_entries Number of entries in eddbuf (below)
261EA/001 ALL edd_mbr_sig_buf_entries Number of entries in edd_mbr_sig_buffer
27 (below)
28290/040 ALL edd_mbr_sig_buffer EDD MBR signatures
292D0/A00 ALL e820_map E820 memory map table
30 (array of struct e820entry)
31D00/1EC ALL eddbuf EDD data (array of struct edd_info)
diff --git a/Documentation/x86/x86_64/00-INDEX b/Documentation/x86/x86_64/00-INDEX
new file mode 100644
index 000000000000..92fc20ab5f0e
--- /dev/null
+++ b/Documentation/x86/x86_64/00-INDEX
@@ -0,0 +1,16 @@
100-INDEX
2 - This file
3boot-options.txt
4 - AMD64-specific boot options.
5cpu-hotplug-spec
6 - Firmware support for CPU hotplug under Linux/x86-64
7fake-numa-for-cpusets
8 - Using numa=fake and CPUSets for Resource Management
9kernel-stacks
10 - Context-specific per-processor interrupt stacks.
11machinecheck
12 - Configurable sysfs parameters for the x86-64 machine check code.
13mm.txt
14 - Memory layout of x86-64 (4 level page tables, 46 bits physical).
15uefi.txt
16 - Booting Linux via Unified Extensible Firmware Interface.
diff --git a/Documentation/x86/x86_64/boot-options.txt b/Documentation/x86/x86_64/boot-options.txt
new file mode 100644
index 000000000000..b0c7b6c4abda
--- /dev/null
+++ b/Documentation/x86/x86_64/boot-options.txt
@@ -0,0 +1,314 @@
1AMD64 specific boot options
2
3There are many others (usually documented in driver documentation), but
4only the AMD64 specific ones are listed here.
5
6Machine check
7
8 mce=off disable machine check
9 mce=bootlog Enable logging of machine checks left over from booting.
10 Disabled by default on AMD because some BIOS leave bogus ones.
11 If your BIOS doesn't do that it's a good idea to enable though
12 to make sure you log even machine check events that result
13 in a reboot. On Intel systems it is enabled by default.
14 mce=nobootlog
15 Disable boot machine check logging.
16 mce=tolerancelevel (number)
17 0: always panic on uncorrected errors, log corrected errors
18 1: panic or SIGBUS on uncorrected errors, log corrected errors
19 2: SIGBUS or log uncorrected errors, log corrected errors
20 3: never panic or SIGBUS, log all errors (for testing only)
21 Default is 1
22 Can be also set using sysfs which is preferable.
23
24 nomce (for compatibility with i386): same as mce=off
25
26 Everything else is in sysfs now.
27
28APICs
29
30 apic Use IO-APIC. Default
31
32 noapic Don't use the IO-APIC.
33
34 disableapic Don't use the local APIC
35
36 nolapic Don't use the local APIC (alias for i386 compatibility)
37
38 pirq=... See Documentation/i386/IO-APIC.txt
39
40 noapictimer Don't set up the APIC timer
41
42 no_timer_check Don't check the IO-APIC timer. This can work around
43 problems with incorrect timer initialization on some boards.
44
45 apicmaintimer Run time keeping from the local APIC timer instead
46 of using the PIT/HPET interrupt for this. This is useful
47 when the PIT/HPET interrupts are unreliable.
48
49 noapicmaintimer Don't do time keeping using the APIC timer.
50 Useful when this option was auto selected, but doesn't work.
51
52 apicpmtimer
53 Do APIC timer calibration using the pmtimer. Implies
54 apicmaintimer. Useful when your PIT timer is totally
55 broken.
56
57 disable_8254_timer / enable_8254_timer
58 Enable interrupt 0 timer routing over the 8254 in addition to over
59 the IO-APIC. The kernel tries to set a sensible default.
60
61Early Console
62
63 syntax: earlyprintk=vga
64 earlyprintk=serial[,ttySn[,baudrate]]
65
66 The early console is useful when the kernel crashes before the
67 normal console is initialized. It is not enabled by
68 default because it has some cosmetic problems.
69 Append ,keep to not disable it when the real console takes over.
70 Only vga or serial at a time, not both.
71 Currently only ttyS0 and ttyS1 are supported.
72 Interaction with the standard serial driver is not very good.
73 The VGA output is eventually overwritten by the real console.
74
75Timing
76
77 notsc
78 Don't use the CPU time stamp counter to read the wall time.
79 This can be used to work around timing problems on multiprocessor systems
80 with not properly synchronized CPUs.
81
82 report_lost_ticks
83 Report when timer interrupts are lost because some code turned off
84 interrupts for too long.
85
86 nmi_watchdog=NUMBER[,panic]
87 NUMBER can be:
88 0 don't use an NMI watchdog
89 1 use the IO-APIC timer for the NMI watchdog
90 2 use the local APIC for the NMI watchdog using a performance counter. Note
91 This will use one performance counter and the local APIC's performance
92 vector.
93 When panic is specified panic when an NMI watchdog timeout occurs.
94 This is useful when you use a panic=... timeout and need the box
95 quickly up again.
96
97 nohpet
98 Don't use the HPET timer.
99
100Idle loop
101
102 idle=poll
103 Don't do power saving in the idle loop using HLT, but poll for rescheduling
104 event. This will make the CPUs eat a lot more power, but may be useful
105 to get slightly better performance in multiprocessor benchmarks. It also
106 makes some profiling using performance counters more accurate.
107 Please note that on systems with MONITOR/MWAIT support (like Intel EM64T
108 CPUs) this option has no performance advantage over the normal idle loop.
109 It may also interact badly with hyperthreading.
110
111Rebooting
112
113 reboot=b[ios] | t[riple] | k[bd] | a[cpi] | e[fi] [, [w]arm | [c]old]
114 bios Use the CPU reboot vector for warm reset
115 warm Don't set the cold reboot flag
116 cold Set the cold reboot flag
117 triple Force a triple fault (init)
118 kbd Use the keyboard controller. cold reset (default)
119 acpi Use the ACPI RESET_REG in the FADT. If ACPI is not configured or the
120 ACPI reset does not work, the reboot path attempts the reset using
121 the keyboard controller.
122 efi Use efi reset_system runtime service. If EFI is not configured or the
123 EFI reset does not work, the reboot path attempts the reset using
124 the keyboard controller.
125
126 Using warm reset will be much faster especially on big memory
127 systems because the BIOS will not go through the memory check.
128 Disadvantage is that not all hardware will be completely reinitialized
129 on reboot so there may be boot problems on some systems.
130
131 reboot=force
132
133 Don't stop other CPUs on reboot. This can make reboot more reliable
134 in some cases.
135
136Non Executable Mappings
137
138 noexec=on|off
139
140 on Enable(default)
141 off Disable
142
143SMP
144
145 additional_cpus=NUM Allow NUM more CPUs for hotplug
146 (defaults are specified by the BIOS, see Documentation/x86_64/cpu-hotplug-spec)
147
148NUMA
149
150 numa=off Only set up a single NUMA node spanning all memory.
151
152 numa=noacpi Don't parse the SRAT table for NUMA setup
153
154 numa=fake=CMDLINE
155 If a number, fakes CMDLINE nodes and ignores NUMA setup of the
156 actual machine. Otherwise, system memory is configured
157 depending on the sizes and coefficients listed. For example:
158 numa=fake=2*512,1024,4*256,*128
159 gives two 512M nodes, a 1024M node, four 256M nodes, and the
160 rest split into 128M chunks. If the last character of CMDLINE
161 is a *, the remaining memory is divided up equally among its
162 coefficient:
163 numa=fake=2*512,2*
164 gives two 512M nodes and the rest split into two nodes.
165 Otherwise, the remaining system RAM is allocated to an
166 additional node.
167
168 numa=hotadd=percent
169 Only allow hotadd memory to preallocate page structures upto
170 percent of already available memory.
171 numa=hotadd=0 will disable hotadd memory.
172
173ACPI
174
175 acpi=off Don't enable ACPI
176 acpi=ht Use ACPI boot table parsing, but don't enable ACPI
177 interpreter
178 acpi=force Force ACPI on (currently not needed)
179
180 acpi=strict Disable out of spec ACPI workarounds.
181
182 acpi_sci={edge,level,high,low} Set up ACPI SCI interrupt.
183
184 acpi=noirq Don't route interrupts
185
186PCI
187
188 pci=off Don't use PCI
189 pci=conf1 Use conf1 access.
190 pci=conf2 Use conf2 access.
191 pci=rom Assign ROMs.
192 pci=assign-busses Assign busses
193 pci=irqmask=MASK Set PCI interrupt mask to MASK
194 pci=lastbus=NUMBER Scan upto NUMBER busses, no matter what the mptable says.
195 pci=noacpi Don't use ACPI to set up PCI interrupt routing.
196
197IOMMU (input/output memory management unit)
198
199 Currently four x86-64 PCI-DMA mapping implementations exist:
200
201 1. <arch/x86_64/kernel/pci-nommu.c>: use no hardware/software IOMMU at all
202 (e.g. because you have < 3 GB memory).
203 Kernel boot message: "PCI-DMA: Disabling IOMMU"
204
205 2. <arch/x86_64/kernel/pci-gart.c>: AMD GART based hardware IOMMU.
206 Kernel boot message: "PCI-DMA: using GART IOMMU"
207
208 3. <arch/x86_64/kernel/pci-swiotlb.c> : Software IOMMU implementation. Used
209 e.g. if there is no hardware IOMMU in the system and it is need because
210 you have >3GB memory or told the kernel to us it (iommu=soft))
211 Kernel boot message: "PCI-DMA: Using software bounce buffering
212 for IO (SWIOTLB)"
213
214 4. <arch/x86_64/pci-calgary.c> : IBM Calgary hardware IOMMU. Used in IBM
215 pSeries and xSeries servers. This hardware IOMMU supports DMA address
216 mapping with memory protection, etc.
217 Kernel boot message: "PCI-DMA: Using Calgary IOMMU"
218
219 iommu=[<size>][,noagp][,off][,force][,noforce][,leak[=<nr_of_leak_pages>]
220 [,memaper[=<order>]][,merge][,forcesac][,fullflush][,nomerge]
221 [,noaperture][,calgary]
222
223 General iommu options:
224 off Don't initialize and use any kind of IOMMU.
225 noforce Don't force hardware IOMMU usage when it is not needed.
226 (default).
227 force Force the use of the hardware IOMMU even when it is
228 not actually needed (e.g. because < 3 GB memory).
229 soft Use software bounce buffering (SWIOTLB) (default for
230 Intel machines). This can be used to prevent the usage
231 of an available hardware IOMMU.
232
233 iommu options only relevant to the AMD GART hardware IOMMU:
234 <size> Set the size of the remapping area in bytes.
235 allowed Overwrite iommu off workarounds for specific chipsets.
236 fullflush Flush IOMMU on each allocation (default).
237 nofullflush Don't use IOMMU fullflush.
238 leak Turn on simple iommu leak tracing (only when
239 CONFIG_IOMMU_LEAK is on). Default number of leak pages
240 is 20.
241 memaper[=<order>] Allocate an own aperture over RAM with size 32MB<<order.
242 (default: order=1, i.e. 64MB)
243 merge Do scatter-gather (SG) merging. Implies "force"
244 (experimental).
245 nomerge Don't do scatter-gather (SG) merging.
246 noaperture Ask the IOMMU not to touch the aperture for AGP.
247 forcesac Force single-address cycle (SAC) mode for masks <40bits
248 (experimental).
249 noagp Don't initialize the AGP driver and use full aperture.
250 allowdac Allow double-address cycle (DAC) mode, i.e. DMA >4GB.
251 DAC is used with 32-bit PCI to push a 64-bit address in
252 two cycles. When off all DMA over >4GB is forced through
253 an IOMMU or software bounce buffering.
254 nodac Forbid DAC mode, i.e. DMA >4GB.
255 panic Always panic when IOMMU overflows.
256 calgary Use the Calgary IOMMU if it is available
257
258 iommu options only relevant to the software bounce buffering (SWIOTLB) IOMMU
259 implementation:
260 swiotlb=<pages>[,force]
261 <pages> Prereserve that many 128K pages for the software IO
262 bounce buffering.
263 force Force all IO through the software TLB.
264
265 Settings for the IBM Calgary hardware IOMMU currently found in IBM
266 pSeries and xSeries machines:
267
268 calgary=[64k,128k,256k,512k,1M,2M,4M,8M]
269 calgary=[translate_empty_slots]
270 calgary=[disable=<PCI bus number>]
271 panic Always panic when IOMMU overflows
272
273 64k,...,8M - Set the size of each PCI slot's translation table
274 when using the Calgary IOMMU. This is the size of the translation
275 table itself in main memory. The smallest table, 64k, covers an IO
276 space of 32MB; the largest, 8MB table, can cover an IO space of
277 4GB. Normally the kernel will make the right choice by itself.
278
279 translate_empty_slots - Enable translation even on slots that have
280 no devices attached to them, in case a device will be hotplugged
281 in the future.
282
283 disable=<PCI bus number> - Disable translation on a given PHB. For
284 example, the built-in graphics adapter resides on the first bridge
285 (PCI bus number 0); if translation (isolation) is enabled on this
286 bridge, X servers that access the hardware directly from user
287 space might stop working. Use this option if you have devices that
288 are accessed from userspace directly on some PCI host bridge.
289
290Debugging
291
292 oops=panic Always panic on oopses. Default is to just kill the process,
293 but there is a small probability of deadlocking the machine.
294 This will also cause panics on machine check exceptions.
295 Useful together with panic=30 to trigger a reboot.
296
297 kstack=N Print N words from the kernel stack in oops dumps.
298
299 pagefaulttrace Dump all page faults. Only useful for extreme debugging
300 and will create a lot of output.
301
302 call_trace=[old|both|newfallback|new]
303 old: use old inexact backtracer
304 new: use new exact dwarf2 unwinder
305 both: print entries from both
306 newfallback: use new unwinder but fall back to old if it gets
307 stuck (default)
308
309Miscellaneous
310
311 nogbpages
312 Do not use GB pages for kernel direct mappings.
313 gbpages
314 Use GB pages for kernel direct mappings.
diff --git a/Documentation/x86/x86_64/cpu-hotplug-spec b/Documentation/x86/x86_64/cpu-hotplug-spec
new file mode 100644
index 000000000000..3c23e0587db3
--- /dev/null
+++ b/Documentation/x86/x86_64/cpu-hotplug-spec
@@ -0,0 +1,21 @@
1Firmware support for CPU hotplug under Linux/x86-64
2---------------------------------------------------
3
4Linux/x86-64 supports CPU hotplug now. For various reasons Linux wants to
5know in advance of boot time the maximum number of CPUs that could be plugged
6into the system. ACPI 3.0 currently has no official way to supply
7this information from the firmware to the operating system.
8
9In ACPI each CPU needs an LAPIC object in the MADT table (5.2.11.5 in the
10ACPI 3.0 specification). ACPI already has the concept of disabled LAPIC
11objects by setting the Enabled bit in the LAPIC object to zero.
12
13For CPU hotplug Linux/x86-64 expects now that any possible future hotpluggable
14CPU is already available in the MADT. If the CPU is not available yet
15it should have its LAPIC Enabled bit set to 0. Linux will use the number
16of disabled LAPICs to compute the maximum number of future CPUs.
17
18In the worst case the user can overwrite this choice using a command line
19option (additional_cpus=...), but it is recommended to supply the correct
20number (or a reasonable approximation of it, with erring towards more not less)
21in the MADT to avoid manual configuration.
diff --git a/Documentation/x86/x86_64/fake-numa-for-cpusets b/Documentation/x86/x86_64/fake-numa-for-cpusets
new file mode 100644
index 000000000000..d1a985c5b00a
--- /dev/null
+++ b/Documentation/x86/x86_64/fake-numa-for-cpusets
@@ -0,0 +1,66 @@
1Using numa=fake and CPUSets for Resource Management
2Written by David Rientjes <rientjes@cs.washington.edu>
3
4This document describes how the numa=fake x86_64 command-line option can be used
5in conjunction with cpusets for coarse memory management. Using this feature,
6you can create fake NUMA nodes that represent contiguous chunks of memory and
7assign them to cpusets and their attached tasks. This is a way of limiting the
8amount of system memory that are available to a certain class of tasks.
9
10For more information on the features of cpusets, see Documentation/cpusets.txt.
11There are a number of different configurations you can use for your needs. For
12more information on the numa=fake command line option and its various ways of
13configuring fake nodes, see Documentation/x86_64/boot-options.txt.
14
15For the purposes of this introduction, we'll assume a very primitive NUMA
16emulation setup of "numa=fake=4*512,". This will split our system memory into
17four equal chunks of 512M each that we can now use to assign to cpusets. As
18you become more familiar with using this combination for resource control,
19you'll determine a better setup to minimize the number of nodes you have to deal
20with.
21
22A machine may be split as follows with "numa=fake=4*512," as reported by dmesg:
23
24 Faking node 0 at 0000000000000000-0000000020000000 (512MB)
25 Faking node 1 at 0000000020000000-0000000040000000 (512MB)
26 Faking node 2 at 0000000040000000-0000000060000000 (512MB)
27 Faking node 3 at 0000000060000000-0000000080000000 (512MB)
28 ...
29 On node 0 totalpages: 130975
30 On node 1 totalpages: 131072
31 On node 2 totalpages: 131072
32 On node 3 totalpages: 131072
33
34Now following the instructions for mounting the cpusets filesystem from
35Documentation/cpusets.txt, you can assign fake nodes (i.e. contiguous memory
36address spaces) to individual cpusets:
37
38 [root@xroads /]# mkdir exampleset
39 [root@xroads /]# mount -t cpuset none exampleset
40 [root@xroads /]# mkdir exampleset/ddset
41 [root@xroads /]# cd exampleset/ddset
42 [root@xroads /exampleset/ddset]# echo 0-1 > cpus
43 [root@xroads /exampleset/ddset]# echo 0-1 > mems
44
45Now this cpuset, 'ddset', will only allowed access to fake nodes 0 and 1 for
46memory allocations (1G).
47
48You can now assign tasks to these cpusets to limit the memory resources
49available to them according to the fake nodes assigned as mems:
50
51 [root@xroads /exampleset/ddset]# echo $$ > tasks
52 [root@xroads /exampleset/ddset]# dd if=/dev/zero of=tmp bs=1024 count=1G
53 [1] 13425
54
55Notice the difference between the system memory usage as reported by
56/proc/meminfo between the restricted cpuset case above and the unrestricted
57case (i.e. running the same 'dd' command without assigning it to a fake NUMA
58cpuset):
59 Unrestricted Restricted
60 MemTotal: 3091900 kB 3091900 kB
61 MemFree: 42113 kB 1513236 kB
62
63This allows for coarse memory management for the tasks you assign to particular
64cpusets. Since cpusets can form a hierarchy, you can create some pretty
65interesting combinations of use-cases for various classes of tasks for your
66memory management needs.
diff --git a/Documentation/x86/x86_64/kernel-stacks b/Documentation/x86/x86_64/kernel-stacks
new file mode 100644
index 000000000000..5ad65d51fb95
--- /dev/null
+++ b/Documentation/x86/x86_64/kernel-stacks
@@ -0,0 +1,99 @@
1Most of the text from Keith Owens, hacked by AK
2
3x86_64 page size (PAGE_SIZE) is 4K.
4
5Like all other architectures, x86_64 has a kernel stack for every
6active thread. These thread stacks are THREAD_SIZE (2*PAGE_SIZE) big.
7These stacks contain useful data as long as a thread is alive or a
8zombie. While the thread is in user space the kernel stack is empty
9except for the thread_info structure at the bottom.
10
11In addition to the per thread stacks, there are specialized stacks
12associated with each CPU. These stacks are only used while the kernel
13is in control on that CPU; when a CPU returns to user space the
14specialized stacks contain no useful data. The main CPU stacks are:
15
16* Interrupt stack. IRQSTACKSIZE
17
18 Used for external hardware interrupts. If this is the first external
19 hardware interrupt (i.e. not a nested hardware interrupt) then the
20 kernel switches from the current task to the interrupt stack. Like
21 the split thread and interrupt stacks on i386 (with CONFIG_4KSTACKS),
22 this gives more room for kernel interrupt processing without having
23 to increase the size of every per thread stack.
24
25 The interrupt stack is also used when processing a softirq.
26
27Switching to the kernel interrupt stack is done by software based on a
28per CPU interrupt nest counter. This is needed because x86-64 "IST"
29hardware stacks cannot nest without races.
30
31x86_64 also has a feature which is not available on i386, the ability
32to automatically switch to a new stack for designated events such as
33double fault or NMI, which makes it easier to handle these unusual
34events on x86_64. This feature is called the Interrupt Stack Table
35(IST). There can be up to 7 IST entries per CPU. The IST code is an
36index into the Task State Segment (TSS). The IST entries in the TSS
37point to dedicated stacks; each stack can be a different size.
38
39An IST is selected by a non-zero value in the IST field of an
40interrupt-gate descriptor. When an interrupt occurs and the hardware
41loads such a descriptor, the hardware automatically sets the new stack
42pointer based on the IST value, then invokes the interrupt handler. If
43software wants to allow nested IST interrupts then the handler must
44adjust the IST values on entry to and exit from the interrupt handler.
45(This is occasionally done, e.g. for debug exceptions.)
46
47Events with different IST codes (i.e. with different stacks) can be
48nested. For example, a debug interrupt can safely be interrupted by an
49NMI. arch/x86_64/kernel/entry.S::paranoidentry adjusts the stack
50pointers on entry to and exit from all IST events, in theory allowing
51IST events with the same code to be nested. However in most cases, the
52stack size allocated to an IST assumes no nesting for the same code.
53If that assumption is ever broken then the stacks will become corrupt.
54
55The currently assigned IST stacks are :-
56
57* STACKFAULT_STACK. EXCEPTION_STKSZ (PAGE_SIZE).
58
59 Used for interrupt 12 - Stack Fault Exception (#SS).
60
61 This allows the CPU to recover from invalid stack segments. Rarely
62 happens.
63
64* DOUBLEFAULT_STACK. EXCEPTION_STKSZ (PAGE_SIZE).
65
66 Used for interrupt 8 - Double Fault Exception (#DF).
67
68 Invoked when handling one exception causes another exception. Happens
69 when the kernel is very confused (e.g. kernel stack pointer corrupt).
70 Using a separate stack allows the kernel to recover from it well enough
71 in many cases to still output an oops.
72
73* NMI_STACK. EXCEPTION_STKSZ (PAGE_SIZE).
74
75 Used for non-maskable interrupts (NMI).
76
77 NMI can be delivered at any time, including when the kernel is in the
78 middle of switching stacks. Using IST for NMI events avoids making
79 assumptions about the previous state of the kernel stack.
80
81* DEBUG_STACK. DEBUG_STKSZ
82
83 Used for hardware debug interrupts (interrupt 1) and for software
84 debug interrupts (INT3).
85
86 When debugging a kernel, debug interrupts (both hardware and
87 software) can occur at any time. Using IST for these interrupts
88 avoids making assumptions about the previous state of the kernel
89 stack.
90
91* MCE_STACK. EXCEPTION_STKSZ (PAGE_SIZE).
92
93 Used for interrupt 18 - Machine Check Exception (#MC).
94
95 MCE can be delivered at any time, including when the kernel is in the
96 middle of switching stacks. Using IST for MCE events avoids making
97 assumptions about the previous state of the kernel stack.
98
99For more details see the Intel IA32 or AMD AMD64 architecture manuals.
diff --git a/Documentation/x86/x86_64/machinecheck b/Documentation/x86/x86_64/machinecheck
new file mode 100644
index 000000000000..a05e58e7b159
--- /dev/null
+++ b/Documentation/x86/x86_64/machinecheck
@@ -0,0 +1,77 @@
1
2Configurable sysfs parameters for the x86-64 machine check code.
3
4Machine checks report internal hardware error conditions detected
5by the CPU. Uncorrected errors typically cause a machine check
6(often with panic), corrected ones cause a machine check log entry.
7
8Machine checks are organized in banks (normally associated with
9a hardware subsystem) and subevents in a bank. The exact meaning
10of the banks and subevent is CPU specific.
11
12mcelog knows how to decode them.
13
14When you see the "Machine check errors logged" message in the system
15log then mcelog should run to collect and decode machine check entries
16from /dev/mcelog. Normally mcelog should be run regularly from a cronjob.
17
18Each CPU has a directory in /sys/devices/system/machinecheck/machinecheckN
19(N = CPU number)
20
21The directory contains some configurable entries:
22
23Entries:
24
25bankNctl
26(N bank number)
27 64bit Hex bitmask enabling/disabling specific subevents for bank N
28 When a bit in the bitmask is zero then the respective
29 subevent will not be reported.
30 By default all events are enabled.
31 Note that BIOS maintain another mask to disable specific events
32 per bank. This is not visible here
33
34The following entries appear for each CPU, but they are truly shared
35between all CPUs.
36
37check_interval
38 How often to poll for corrected machine check errors, in seconds
39 (Note output is hexademical). Default 5 minutes. When the poller
40 finds MCEs it triggers an exponential speedup (poll more often) on
41 the polling interval. When the poller stops finding MCEs, it
42 triggers an exponential backoff (poll less often) on the polling
43 interval. The check_interval variable is both the initial and
44 maximum polling interval.
45
46tolerant
47 Tolerance level. When a machine check exception occurs for a non
48 corrected machine check the kernel can take different actions.
49 Since machine check exceptions can happen any time it is sometimes
50 risky for the kernel to kill a process because it defies
51 normal kernel locking rules. The tolerance level configures
52 how hard the kernel tries to recover even at some risk of
53 deadlock. Higher tolerant values trade potentially better uptime
54 with the risk of a crash or even corruption (for tolerant >= 3).
55
56 0: always panic on uncorrected errors, log corrected errors
57 1: panic or SIGBUS on uncorrected errors, log corrected errors
58 2: SIGBUS or log uncorrected errors, log corrected errors
59 3: never panic or SIGBUS, log all errors (for testing only)
60
61 Default: 1
62
63 Note this only makes a difference if the CPU allows recovery
64 from a machine check exception. Current x86 CPUs generally do not.
65
66trigger
67 Program to run when a machine check event is detected.
68 This is an alternative to running mcelog regularly from cron
69 and allows to detect events faster.
70
71TBD document entries for AMD threshold interrupt configuration
72
73For more details about the x86 machine check architecture
74see the Intel and AMD architecture manuals from their developer websites.
75
76For more details about the architecture see
77see http://one.firstfloor.org/~andi/mce.pdf
diff --git a/Documentation/x86/x86_64/mm.txt b/Documentation/x86/x86_64/mm.txt
new file mode 100644
index 000000000000..efce75097369
--- /dev/null
+++ b/Documentation/x86/x86_64/mm.txt
@@ -0,0 +1,28 @@
1
2<previous description obsolete, deleted>
3
4Virtual memory map with 4 level page tables:
5
60000000000000000 - 00007fffffffffff (=47 bits) user space, different per mm
7hole caused by [48:63] sign extension
8ffff800000000000 - ffff80ffffffffff (=40 bits) guard hole
9ffff810000000000 - ffffc0ffffffffff (=46 bits) direct mapping of all phys. memory
10ffffc10000000000 - ffffc1ffffffffff (=40 bits) hole
11ffffc20000000000 - ffffe1ffffffffff (=45 bits) vmalloc/ioremap space
12ffffe20000000000 - ffffe2ffffffffff (=40 bits) virtual memory map (1TB)
13... unused hole ...
14ffffffff80000000 - ffffffffa0000000 (=512 MB) kernel text mapping, from phys 0
15ffffffffa0000000 - fffffffffff00000 (=1536 MB) module mapping space
16
17The direct mapping covers all memory in the system up to the highest
18memory address (this means in some cases it can also include PCI memory
19holes).
20
21vmalloc space is lazily synchronized into the different PML4 pages of
22the processes using the page fault handler, with init_level4_pgt as
23reference.
24
25Current X86-64 implementations only support 40 bits of address space,
26but we support up to 46 bits. This expands into MBZ space in the page tables.
27
28-Andi Kleen, Jul 2004
diff --git a/Documentation/x86/x86_64/uefi.txt b/Documentation/x86/x86_64/uefi.txt
new file mode 100644
index 000000000000..7d77120a5184
--- /dev/null
+++ b/Documentation/x86/x86_64/uefi.txt
@@ -0,0 +1,38 @@
1General note on [U]EFI x86_64 support
2-------------------------------------
3
4The nomenclature EFI and UEFI are used interchangeably in this document.
5
6Although the tools below are _not_ needed for building the kernel,
7the needed bootloader support and associated tools for x86_64 platforms
8with EFI firmware and specifications are listed below.
9
101. UEFI specification: http://www.uefi.org
11
122. Booting Linux kernel on UEFI x86_64 platform requires bootloader
13 support. Elilo with x86_64 support can be used.
14
153. x86_64 platform with EFI/UEFI firmware.
16
17Mechanics:
18---------
19- Build the kernel with the following configuration.
20 CONFIG_FB_EFI=y
21 CONFIG_FRAMEBUFFER_CONSOLE=y
22 If EFI runtime services are expected, the following configuration should
23 be selected.
24 CONFIG_EFI=y
25 CONFIG_EFI_VARS=y or m # optional
26- Create a VFAT partition on the disk
27- Copy the following to the VFAT partition:
28 elilo bootloader with x86_64 support, elilo configuration file,
29 kernel image built in first step and corresponding
30 initrd. Instructions on building elilo and its dependencies
31 can be found in the elilo sourceforge project.
32- Boot to EFI shell and invoke elilo choosing the kernel image built
33 in first step.
34- If some or all EFI runtime services don't work, you can try following
35 kernel command line parameters to turn off some or all EFI runtime
36 services.
37 noefi turn off all EFI runtime services
38 reboot_type=k turn off EFI reboot runtime service