debuggers.hg

view xen/include/public/xen.h @ 19826:2f9e1348aa98

x86_64: allow more vCPU-s per guest

Since the shared info layout is fixed, guests are required to use
VCPUOP_register_vcpu_info prior to booting any vCPU beyond the
traditional limit of 32.

MAX_VIRT_CPUS, being an implemetation detail of the hypervisor, is no
longer being exposed in the public headers.

The tools changes are clearly incomplete (and done only so things
would
build again), and the current state of the tools (using scalar
variables all over the place to represent vCPU bitmaps) very likely
doesn't permit booting DomU-s with more than the traditional number of
vCPU-s. Testing of the extended functionality was done with Dom0 (96
vCPU-s, as well as 128 vCPU-s out of which the kernel elected - by way
of a simple kernel side patch - to use only some, resulting in a
sparse
bitmap).

ia64 changes only to make things build, and build-tested only (and the
tools part only as far as the build would go without encountering
unrelated problems in the blktap code).

Signed-off-by: Jan Beulich <jbeulich@novell.com>
author Keir Fraser <keir.fraser@citrix.com>
date Thu Jun 18 10:14:16 2009 +0100 (2009-06-18)
parents f210a633571c
children 01ae7dc043ba
line source
1 /******************************************************************************
2 * xen.h
3 *
4 * Guest OS interface to Xen.
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to
8 * deal in the Software without restriction, including without limitation the
9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10 * sell copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
12 *
13 * The above copyright notice and this permission notice shall be included in
14 * all copies or substantial portions of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
21 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
22 * DEALINGS IN THE SOFTWARE.
23 *
24 * Copyright (c) 2004, K A Fraser
25 */
27 #ifndef __XEN_PUBLIC_XEN_H__
28 #define __XEN_PUBLIC_XEN_H__
30 #include "xen-compat.h"
32 #if defined(__i386__) || defined(__x86_64__)
33 #include "arch-x86/xen.h"
34 #elif defined(__ia64__)
35 #include "arch-ia64.h"
36 #else
37 #error "Unsupported architecture"
38 #endif
40 #ifndef __ASSEMBLY__
41 /* Guest handles for primitive C types. */
42 DEFINE_XEN_GUEST_HANDLE(char);
43 __DEFINE_XEN_GUEST_HANDLE(uchar, unsigned char);
44 DEFINE_XEN_GUEST_HANDLE(int);
45 __DEFINE_XEN_GUEST_HANDLE(uint, unsigned int);
46 DEFINE_XEN_GUEST_HANDLE(long);
47 __DEFINE_XEN_GUEST_HANDLE(ulong, unsigned long);
48 DEFINE_XEN_GUEST_HANDLE(void);
50 DEFINE_XEN_GUEST_HANDLE(xen_pfn_t);
51 #endif
53 /*
54 * HYPERCALLS
55 */
57 #define __HYPERVISOR_set_trap_table 0
58 #define __HYPERVISOR_mmu_update 1
59 #define __HYPERVISOR_set_gdt 2
60 #define __HYPERVISOR_stack_switch 3
61 #define __HYPERVISOR_set_callbacks 4
62 #define __HYPERVISOR_fpu_taskswitch 5
63 #define __HYPERVISOR_sched_op_compat 6 /* compat since 0x00030101 */
64 #define __HYPERVISOR_platform_op 7
65 #define __HYPERVISOR_set_debugreg 8
66 #define __HYPERVISOR_get_debugreg 9
67 #define __HYPERVISOR_update_descriptor 10
68 #define __HYPERVISOR_memory_op 12
69 #define __HYPERVISOR_multicall 13
70 #define __HYPERVISOR_update_va_mapping 14
71 #define __HYPERVISOR_set_timer_op 15
72 #define __HYPERVISOR_event_channel_op_compat 16 /* compat since 0x00030202 */
73 #define __HYPERVISOR_xen_version 17
74 #define __HYPERVISOR_console_io 18
75 #define __HYPERVISOR_physdev_op_compat 19 /* compat since 0x00030202 */
76 #define __HYPERVISOR_grant_table_op 20
77 #define __HYPERVISOR_vm_assist 21
78 #define __HYPERVISOR_update_va_mapping_otherdomain 22
79 #define __HYPERVISOR_iret 23 /* x86 only */
80 #define __HYPERVISOR_vcpu_op 24
81 #define __HYPERVISOR_set_segment_base 25 /* x86/64 only */
82 #define __HYPERVISOR_mmuext_op 26
83 #define __HYPERVISOR_xsm_op 27
84 #define __HYPERVISOR_nmi_op 28
85 #define __HYPERVISOR_sched_op 29
86 #define __HYPERVISOR_callback_op 30
87 #define __HYPERVISOR_xenoprof_op 31
88 #define __HYPERVISOR_event_channel_op 32
89 #define __HYPERVISOR_physdev_op 33
90 #define __HYPERVISOR_hvm_op 34
91 #define __HYPERVISOR_sysctl 35
92 #define __HYPERVISOR_domctl 36
93 #define __HYPERVISOR_kexec_op 37
94 #define __HYPERVISOR_tmem_op 38
96 /* Architecture-specific hypercall definitions. */
97 #define __HYPERVISOR_arch_0 48
98 #define __HYPERVISOR_arch_1 49
99 #define __HYPERVISOR_arch_2 50
100 #define __HYPERVISOR_arch_3 51
101 #define __HYPERVISOR_arch_4 52
102 #define __HYPERVISOR_arch_5 53
103 #define __HYPERVISOR_arch_6 54
104 #define __HYPERVISOR_arch_7 55
106 /*
107 * HYPERCALL COMPATIBILITY.
108 */
110 /* New sched_op hypercall introduced in 0x00030101. */
111 #if __XEN_INTERFACE_VERSION__ < 0x00030101
112 #undef __HYPERVISOR_sched_op
113 #define __HYPERVISOR_sched_op __HYPERVISOR_sched_op_compat
114 #endif
116 /* New event-channel and physdev hypercalls introduced in 0x00030202. */
117 #if __XEN_INTERFACE_VERSION__ < 0x00030202
118 #undef __HYPERVISOR_event_channel_op
119 #define __HYPERVISOR_event_channel_op __HYPERVISOR_event_channel_op_compat
120 #undef __HYPERVISOR_physdev_op
121 #define __HYPERVISOR_physdev_op __HYPERVISOR_physdev_op_compat
122 #endif
124 /* New platform_op hypercall introduced in 0x00030204. */
125 #if __XEN_INTERFACE_VERSION__ < 0x00030204
126 #define __HYPERVISOR_dom0_op __HYPERVISOR_platform_op
127 #endif
129 /*
130 * VIRTUAL INTERRUPTS
131 *
132 * Virtual interrupts that a guest OS may receive from Xen.
133 *
134 * In the side comments, 'V.' denotes a per-VCPU VIRQ while 'G.' denotes a
135 * global VIRQ. The former can be bound once per VCPU and cannot be re-bound.
136 * The latter can be allocated only once per guest: they must initially be
137 * allocated to VCPU0 but can subsequently be re-bound.
138 */
139 #define VIRQ_TIMER 0 /* V. Timebase update, and/or requested timeout. */
140 #define VIRQ_DEBUG 1 /* V. Request guest to dump debug info. */
141 #define VIRQ_CONSOLE 2 /* G. (DOM0) Bytes received on emergency console. */
142 #define VIRQ_DOM_EXC 3 /* G. (DOM0) Exceptional event for some domain. */
143 #define VIRQ_TBUF 4 /* G. (DOM0) Trace buffer has records available. */
144 #define VIRQ_DEBUGGER 6 /* G. (DOM0) A domain has paused for debugging. */
145 #define VIRQ_XENOPROF 7 /* V. XenOprofile interrupt: new sample available */
146 #define VIRQ_CON_RING 8 /* G. (DOM0) Bytes received on console */
148 /* Architecture-specific VIRQ definitions. */
149 #define VIRQ_ARCH_0 16
150 #define VIRQ_ARCH_1 17
151 #define VIRQ_ARCH_2 18
152 #define VIRQ_ARCH_3 19
153 #define VIRQ_ARCH_4 20
154 #define VIRQ_ARCH_5 21
155 #define VIRQ_ARCH_6 22
156 #define VIRQ_ARCH_7 23
158 #define NR_VIRQS 24
160 /*
161 * MMU-UPDATE REQUESTS
162 *
163 * HYPERVISOR_mmu_update() accepts a list of (ptr, val) pairs.
164 * A foreigndom (FD) can be specified (or DOMID_SELF for none).
165 * Where the FD has some effect, it is described below.
166 * ptr[1:0] specifies the appropriate MMU_* command.
167 *
168 * ptr[1:0] == MMU_NORMAL_PT_UPDATE:
169 * Updates an entry in a page table. If updating an L1 table, and the new
170 * table entry is valid/present, the mapped frame must belong to the FD, if
171 * an FD has been specified. If attempting to map an I/O page then the
172 * caller assumes the privilege of the FD.
173 * FD == DOMID_IO: Permit /only/ I/O mappings, at the priv level of the caller.
174 * FD == DOMID_XEN: Map restricted areas of Xen's heap space.
175 * ptr[:2] -- Machine address of the page-table entry to modify.
176 * val -- Value to write.
177 *
178 * ptr[1:0] == MMU_MACHPHYS_UPDATE:
179 * Updates an entry in the machine->pseudo-physical mapping table.
180 * ptr[:2] -- Machine address within the frame whose mapping to modify.
181 * The frame must belong to the FD, if one is specified.
182 * val -- Value to write into the mapping entry.
183 *
184 * ptr[1:0] == MMU_PT_UPDATE_PRESERVE_AD:
185 * As MMU_NORMAL_PT_UPDATE above, but A/D bits currently in the PTE are ORed
186 * with those in @val.
187 */
188 #define MMU_NORMAL_PT_UPDATE 0 /* checked '*ptr = val'. ptr is MA. */
189 #define MMU_MACHPHYS_UPDATE 1 /* ptr = MA of frame to modify entry for */
190 #define MMU_PT_UPDATE_PRESERVE_AD 2 /* atomically: *ptr = val | (*ptr&(A|D)) */
192 /*
193 * MMU EXTENDED OPERATIONS
194 *
195 * HYPERVISOR_mmuext_op() accepts a list of mmuext_op structures.
196 * A foreigndom (FD) can be specified (or DOMID_SELF for none).
197 * Where the FD has some effect, it is described below.
198 *
199 * cmd: MMUEXT_(UN)PIN_*_TABLE
200 * mfn: Machine frame number to be (un)pinned as a p.t. page.
201 * The frame must belong to the FD, if one is specified.
202 *
203 * cmd: MMUEXT_NEW_BASEPTR
204 * mfn: Machine frame number of new page-table base to install in MMU.
205 *
206 * cmd: MMUEXT_NEW_USER_BASEPTR [x86/64 only]
207 * mfn: Machine frame number of new page-table base to install in MMU
208 * when in user space.
209 *
210 * cmd: MMUEXT_TLB_FLUSH_LOCAL
211 * No additional arguments. Flushes local TLB.
212 *
213 * cmd: MMUEXT_INVLPG_LOCAL
214 * linear_addr: Linear address to be flushed from the local TLB.
215 *
216 * cmd: MMUEXT_TLB_FLUSH_MULTI
217 * vcpumask: Pointer to bitmap of VCPUs to be flushed.
218 *
219 * cmd: MMUEXT_INVLPG_MULTI
220 * linear_addr: Linear address to be flushed.
221 * vcpumask: Pointer to bitmap of VCPUs to be flushed.
222 *
223 * cmd: MMUEXT_TLB_FLUSH_ALL
224 * No additional arguments. Flushes all VCPUs' TLBs.
225 *
226 * cmd: MMUEXT_INVLPG_ALL
227 * linear_addr: Linear address to be flushed from all VCPUs' TLBs.
228 *
229 * cmd: MMUEXT_FLUSH_CACHE
230 * No additional arguments. Writes back and flushes cache contents.
231 *
232 * cmd: MMUEXT_SET_LDT
233 * linear_addr: Linear address of LDT base (NB. must be page-aligned).
234 * nr_ents: Number of entries in LDT.
235 *
236 * cmd: MMUEXT_CLEAR_PAGE
237 * mfn: Machine frame number to be cleared.
238 *
239 * cmd: MMUEXT_COPY_PAGE
240 * mfn: Machine frame number of the destination page.
241 * src_mfn: Machine frame number of the source page.
242 */
243 #define MMUEXT_PIN_L1_TABLE 0
244 #define MMUEXT_PIN_L2_TABLE 1
245 #define MMUEXT_PIN_L3_TABLE 2
246 #define MMUEXT_PIN_L4_TABLE 3
247 #define MMUEXT_UNPIN_TABLE 4
248 #define MMUEXT_NEW_BASEPTR 5
249 #define MMUEXT_TLB_FLUSH_LOCAL 6
250 #define MMUEXT_INVLPG_LOCAL 7
251 #define MMUEXT_TLB_FLUSH_MULTI 8
252 #define MMUEXT_INVLPG_MULTI 9
253 #define MMUEXT_TLB_FLUSH_ALL 10
254 #define MMUEXT_INVLPG_ALL 11
255 #define MMUEXT_FLUSH_CACHE 12
256 #define MMUEXT_SET_LDT 13
257 #define MMUEXT_NEW_USER_BASEPTR 15
258 #define MMUEXT_CLEAR_PAGE 16
259 #define MMUEXT_COPY_PAGE 17
261 #ifndef __ASSEMBLY__
262 struct mmuext_op {
263 unsigned int cmd;
264 union {
265 /* [UN]PIN_TABLE, NEW_BASEPTR, NEW_USER_BASEPTR
266 * CLEAR_PAGE, COPY_PAGE */
267 xen_pfn_t mfn;
268 /* INVLPG_LOCAL, INVLPG_ALL, SET_LDT */
269 unsigned long linear_addr;
270 } arg1;
271 union {
272 /* SET_LDT */
273 unsigned int nr_ents;
274 /* TLB_FLUSH_MULTI, INVLPG_MULTI */
275 #if __XEN_INTERFACE_VERSION__ >= 0x00030205
276 XEN_GUEST_HANDLE(const_void) vcpumask;
277 #else
278 const void *vcpumask;
279 #endif
280 /* COPY_PAGE */
281 xen_pfn_t src_mfn;
282 } arg2;
283 };
284 typedef struct mmuext_op mmuext_op_t;
285 DEFINE_XEN_GUEST_HANDLE(mmuext_op_t);
286 #endif
288 /* These are passed as 'flags' to update_va_mapping. They can be ORed. */
289 /* When specifying UVMF_MULTI, also OR in a pointer to a CPU bitmap. */
290 /* UVMF_LOCAL is merely UVMF_MULTI with a NULL bitmap pointer. */
291 #define UVMF_NONE (0UL<<0) /* No flushing at all. */
292 #define UVMF_TLB_FLUSH (1UL<<0) /* Flush entire TLB(s). */
293 #define UVMF_INVLPG (2UL<<0) /* Flush only one entry. */
294 #define UVMF_FLUSHTYPE_MASK (3UL<<0)
295 #define UVMF_MULTI (0UL<<2) /* Flush subset of TLBs. */
296 #define UVMF_LOCAL (0UL<<2) /* Flush local TLB. */
297 #define UVMF_ALL (1UL<<2) /* Flush all TLBs. */
299 /*
300 * Commands to HYPERVISOR_console_io().
301 */
302 #define CONSOLEIO_write 0
303 #define CONSOLEIO_read 1
305 /*
306 * Commands to HYPERVISOR_vm_assist().
307 */
308 #define VMASST_CMD_enable 0
309 #define VMASST_CMD_disable 1
311 /* x86/32 guests: simulate full 4GB segment limits. */
312 #define VMASST_TYPE_4gb_segments 0
314 /* x86/32 guests: trap (vector 15) whenever above vmassist is used. */
315 #define VMASST_TYPE_4gb_segments_notify 1
317 /*
318 * x86 guests: support writes to bottom-level PTEs.
319 * NB1. Page-directory entries cannot be written.
320 * NB2. Guest must continue to remove all writable mappings of PTEs.
321 */
322 #define VMASST_TYPE_writable_pagetables 2
324 /* x86/PAE guests: support PDPTs above 4GB. */
325 #define VMASST_TYPE_pae_extended_cr3 3
327 #define MAX_VMASST_TYPE 3
329 #ifndef __ASSEMBLY__
331 typedef uint16_t domid_t;
333 /* Domain ids >= DOMID_FIRST_RESERVED cannot be used for ordinary domains. */
334 #define DOMID_FIRST_RESERVED (0x7FF0U)
336 /* DOMID_SELF is used in certain contexts to refer to oneself. */
337 #define DOMID_SELF (0x7FF0U)
339 /*
340 * DOMID_IO is used to restrict page-table updates to mapping I/O memory.
341 * Although no Foreign Domain need be specified to map I/O pages, DOMID_IO
342 * is useful to ensure that no mappings to the OS's own heap are accidentally
343 * installed. (e.g., in Linux this could cause havoc as reference counts
344 * aren't adjusted on the I/O-mapping code path).
345 * This only makes sense in MMUEXT_SET_FOREIGNDOM, but in that context can
346 * be specified by any calling domain.
347 */
348 #define DOMID_IO (0x7FF1U)
350 /*
351 * DOMID_XEN is used to allow privileged domains to map restricted parts of
352 * Xen's heap space (e.g., the machine_to_phys table).
353 * This only makes sense in MMUEXT_SET_FOREIGNDOM, and is only permitted if
354 * the caller is privileged.
355 */
356 #define DOMID_XEN (0x7FF2U)
358 /* DOMID_INVALID is used to identity invalid domid */
359 #define DOMID_INVALID (0x7FFFU)
361 /*
362 * Send an array of these to HYPERVISOR_mmu_update().
363 * NB. The fields are natural pointer/address size for this architecture.
364 */
365 struct mmu_update {
366 uint64_t ptr; /* Machine address of PTE. */
367 uint64_t val; /* New contents of PTE. */
368 };
369 typedef struct mmu_update mmu_update_t;
370 DEFINE_XEN_GUEST_HANDLE(mmu_update_t);
372 /*
373 * Send an array of these to HYPERVISOR_multicall().
374 * NB. The fields are natural register size for this architecture.
375 */
376 struct multicall_entry {
377 unsigned long op, result;
378 unsigned long args[6];
379 };
380 typedef struct multicall_entry multicall_entry_t;
381 DEFINE_XEN_GUEST_HANDLE(multicall_entry_t);
383 /*
384 * Event channel endpoints per domain:
385 * 1024 if a long is 32 bits; 4096 if a long is 64 bits.
386 */
387 #define NR_EVENT_CHANNELS (sizeof(unsigned long) * sizeof(unsigned long) * 64)
389 struct vcpu_time_info {
390 /*
391 * Updates to the following values are preceded and followed by an
392 * increment of 'version'. The guest can therefore detect updates by
393 * looking for changes to 'version'. If the least-significant bit of
394 * the version number is set then an update is in progress and the guest
395 * must wait to read a consistent set of values.
396 * The correct way to interact with the version number is similar to
397 * Linux's seqlock: see the implementations of read_seqbegin/read_seqretry.
398 */
399 uint32_t version;
400 uint32_t pad0;
401 uint64_t tsc_timestamp; /* TSC at last update of time vals. */
402 uint64_t system_time; /* Time, in nanosecs, since boot. */
403 /*
404 * Current system time:
405 * system_time +
406 * ((((tsc - tsc_timestamp) << tsc_shift) * tsc_to_system_mul) >> 32)
407 * CPU frequency (Hz):
408 * ((10^9 << 32) / tsc_to_system_mul) >> tsc_shift
409 */
410 uint32_t tsc_to_system_mul;
411 int8_t tsc_shift;
412 int8_t pad1[3];
413 }; /* 32 bytes */
414 typedef struct vcpu_time_info vcpu_time_info_t;
416 struct vcpu_info {
417 /*
418 * 'evtchn_upcall_pending' is written non-zero by Xen to indicate
419 * a pending notification for a particular VCPU. It is then cleared
420 * by the guest OS /before/ checking for pending work, thus avoiding
421 * a set-and-check race. Note that the mask is only accessed by Xen
422 * on the CPU that is currently hosting the VCPU. This means that the
423 * pending and mask flags can be updated by the guest without special
424 * synchronisation (i.e., no need for the x86 LOCK prefix).
425 * This may seem suboptimal because if the pending flag is set by
426 * a different CPU then an IPI may be scheduled even when the mask
427 * is set. However, note:
428 * 1. The task of 'interrupt holdoff' is covered by the per-event-
429 * channel mask bits. A 'noisy' event that is continually being
430 * triggered can be masked at source at this very precise
431 * granularity.
432 * 2. The main purpose of the per-VCPU mask is therefore to restrict
433 * reentrant execution: whether for concurrency control, or to
434 * prevent unbounded stack usage. Whatever the purpose, we expect
435 * that the mask will be asserted only for short periods at a time,
436 * and so the likelihood of a 'spurious' IPI is suitably small.
437 * The mask is read before making an event upcall to the guest: a
438 * non-zero mask therefore guarantees that the VCPU will not receive
439 * an upcall activation. The mask is cleared when the VCPU requests
440 * to block: this avoids wakeup-waiting races.
441 */
442 uint8_t evtchn_upcall_pending;
443 uint8_t evtchn_upcall_mask;
444 unsigned long evtchn_pending_sel;
445 struct arch_vcpu_info arch;
446 struct vcpu_time_info time;
447 }; /* 64 bytes (x86) */
448 #ifndef __XEN__
449 typedef struct vcpu_info vcpu_info_t;
450 #endif
452 /*
453 * Xen/kernel shared data -- pointer provided in start_info.
454 *
455 * This structure is defined to be both smaller than a page, and the
456 * only data on the shared page, but may vary in actual size even within
457 * compatible Xen versions; guests should not rely on the size
458 * of this structure remaining constant.
459 */
460 struct shared_info {
461 struct vcpu_info vcpu_info[XEN_LEGACY_MAX_VCPUS];
463 /*
464 * A domain can create "event channels" on which it can send and receive
465 * asynchronous event notifications. There are three classes of event that
466 * are delivered by this mechanism:
467 * 1. Bi-directional inter- and intra-domain connections. Domains must
468 * arrange out-of-band to set up a connection (usually by allocating
469 * an unbound 'listener' port and avertising that via a storage service
470 * such as xenstore).
471 * 2. Physical interrupts. A domain with suitable hardware-access
472 * privileges can bind an event-channel port to a physical interrupt
473 * source.
474 * 3. Virtual interrupts ('events'). A domain can bind an event-channel
475 * port to a virtual interrupt source, such as the virtual-timer
476 * device or the emergency console.
477 *
478 * Event channels are addressed by a "port index". Each channel is
479 * associated with two bits of information:
480 * 1. PENDING -- notifies the domain that there is a pending notification
481 * to be processed. This bit is cleared by the guest.
482 * 2. MASK -- if this bit is clear then a 0->1 transition of PENDING
483 * will cause an asynchronous upcall to be scheduled. This bit is only
484 * updated by the guest. It is read-only within Xen. If a channel
485 * becomes pending while the channel is masked then the 'edge' is lost
486 * (i.e., when the channel is unmasked, the guest must manually handle
487 * pending notifications as no upcall will be scheduled by Xen).
488 *
489 * To expedite scanning of pending notifications, any 0->1 pending
490 * transition on an unmasked channel causes a corresponding bit in a
491 * per-vcpu selector word to be set. Each bit in the selector covers a
492 * 'C long' in the PENDING bitfield array.
493 */
494 unsigned long evtchn_pending[sizeof(unsigned long) * 8];
495 unsigned long evtchn_mask[sizeof(unsigned long) * 8];
497 /*
498 * Wallclock time: updated only by control software. Guests should base
499 * their gettimeofday() syscall on this wallclock-base value.
500 */
501 uint32_t wc_version; /* Version counter: see vcpu_time_info_t. */
502 uint32_t wc_sec; /* Secs 00:00:00 UTC, Jan 1, 1970. */
503 uint32_t wc_nsec; /* Nsecs 00:00:00 UTC, Jan 1, 1970. */
505 struct arch_shared_info arch;
507 };
508 #ifndef __XEN__
509 typedef struct shared_info shared_info_t;
510 #endif
512 /*
513 * Start-of-day memory layout:
514 * 1. The domain is started within contiguous virtual-memory region.
515 * 2. The contiguous region ends on an aligned 4MB boundary.
516 * 3. This the order of bootstrap elements in the initial virtual region:
517 * a. relocated kernel image
518 * b. initial ram disk [mod_start, mod_len]
519 * c. list of allocated page frames [mfn_list, nr_pages]
520 * (unless relocated due to XEN_ELFNOTE_INIT_P2M)
521 * d. start_info_t structure [register ESI (x86)]
522 * e. bootstrap page tables [pt_base, CR3 (x86)]
523 * f. bootstrap stack [register ESP (x86)]
524 * 4. Bootstrap elements are packed together, but each is 4kB-aligned.
525 * 5. The initial ram disk may be omitted.
526 * 6. The list of page frames forms a contiguous 'pseudo-physical' memory
527 * layout for the domain. In particular, the bootstrap virtual-memory
528 * region is a 1:1 mapping to the first section of the pseudo-physical map.
529 * 7. All bootstrap elements are mapped read-writable for the guest OS. The
530 * only exception is the bootstrap page table, which is mapped read-only.
531 * 8. There is guaranteed to be at least 512kB padding after the final
532 * bootstrap element. If necessary, the bootstrap virtual region is
533 * extended by an extra 4MB to ensure this.
534 */
536 #define MAX_GUEST_CMDLINE 1024
537 struct start_info {
538 /* THE FOLLOWING ARE FILLED IN BOTH ON INITIAL BOOT AND ON RESUME. */
539 char magic[32]; /* "xen-<version>-<platform>". */
540 unsigned long nr_pages; /* Total pages allocated to this domain. */
541 unsigned long shared_info; /* MACHINE address of shared info struct. */
542 uint32_t flags; /* SIF_xxx flags. */
543 xen_pfn_t store_mfn; /* MACHINE page number of shared page. */
544 uint32_t store_evtchn; /* Event channel for store communication. */
545 union {
546 struct {
547 xen_pfn_t mfn; /* MACHINE page number of console page. */
548 uint32_t evtchn; /* Event channel for console page. */
549 } domU;
550 struct {
551 uint32_t info_off; /* Offset of console_info struct. */
552 uint32_t info_size; /* Size of console_info struct from start.*/
553 } dom0;
554 } console;
555 /* THE FOLLOWING ARE ONLY FILLED IN ON INITIAL BOOT (NOT RESUME). */
556 unsigned long pt_base; /* VIRTUAL address of page directory. */
557 unsigned long nr_pt_frames; /* Number of bootstrap p.t. frames. */
558 unsigned long mfn_list; /* VIRTUAL address of page-frame list. */
559 unsigned long mod_start; /* VIRTUAL address of pre-loaded module. */
560 unsigned long mod_len; /* Size (bytes) of pre-loaded module. */
561 int8_t cmd_line[MAX_GUEST_CMDLINE];
562 /* The pfn range here covers both page table and p->m table frames. */
563 unsigned long first_p2m_pfn;/* 1st pfn forming initial P->M table. */
564 unsigned long nr_p2m_frames;/* # of pfns forming initial P->M table. */
565 };
566 typedef struct start_info start_info_t;
568 /* New console union for dom0 introduced in 0x00030203. */
569 #if __XEN_INTERFACE_VERSION__ < 0x00030203
570 #define console_mfn console.domU.mfn
571 #define console_evtchn console.domU.evtchn
572 #endif
574 /* These flags are passed in the 'flags' field of start_info_t. */
575 #define SIF_PRIVILEGED (1<<0) /* Is the domain privileged? */
576 #define SIF_INITDOMAIN (1<<1) /* Is this the initial control domain? */
577 #define SIF_PM_MASK (0xFF<<8) /* reserve 1 byte for xen-pm options */
579 typedef struct dom0_vga_console_info {
580 uint8_t video_type; /* DOM0_VGA_CONSOLE_??? */
581 #define XEN_VGATYPE_TEXT_MODE_3 0x03
582 #define XEN_VGATYPE_VESA_LFB 0x23
584 union {
585 struct {
586 /* Font height, in pixels. */
587 uint16_t font_height;
588 /* Cursor location (column, row). */
589 uint16_t cursor_x, cursor_y;
590 /* Number of rows and columns (dimensions in characters). */
591 uint16_t rows, columns;
592 } text_mode_3;
594 struct {
595 /* Width and height, in pixels. */
596 uint16_t width, height;
597 /* Bytes per scan line. */
598 uint16_t bytes_per_line;
599 /* Bits per pixel. */
600 uint16_t bits_per_pixel;
601 /* LFB physical address, and size (in units of 64kB). */
602 uint32_t lfb_base;
603 uint32_t lfb_size;
604 /* RGB mask offsets and sizes, as defined by VBE 1.2+ */
605 uint8_t red_pos, red_size;
606 uint8_t green_pos, green_size;
607 uint8_t blue_pos, blue_size;
608 uint8_t rsvd_pos, rsvd_size;
609 #if __XEN_INTERFACE_VERSION__ >= 0x00030206
610 /* VESA capabilities (offset 0xa, VESA command 0x4f00). */
611 uint32_t gbl_caps;
612 /* Mode attributes (offset 0x0, VESA command 0x4f01). */
613 uint16_t mode_attrs;
614 #endif
615 } vesa_lfb;
616 } u;
617 } dom0_vga_console_info_t;
618 #define xen_vga_console_info dom0_vga_console_info
619 #define xen_vga_console_info_t dom0_vga_console_info_t
621 typedef uint8_t xen_domain_handle_t[16];
623 /* Turn a plain number into a C unsigned long constant. */
624 #define __mk_unsigned_long(x) x ## UL
625 #define mk_unsigned_long(x) __mk_unsigned_long(x)
627 __DEFINE_XEN_GUEST_HANDLE(uint8, uint8_t);
628 __DEFINE_XEN_GUEST_HANDLE(uint16, uint16_t);
629 __DEFINE_XEN_GUEST_HANDLE(uint32, uint32_t);
630 __DEFINE_XEN_GUEST_HANDLE(uint64, uint64_t);
632 #else /* __ASSEMBLY__ */
634 /* In assembly code we cannot use C numeric constant suffixes. */
635 #define mk_unsigned_long(x) x
637 #endif /* !__ASSEMBLY__ */
639 /* Default definitions for macros used by domctl/sysctl. */
640 #if defined(__XEN__) || defined(__XEN_TOOLS__)
641 #ifndef uint64_aligned_t
642 #define uint64_aligned_t uint64_t
643 #endif
644 #ifndef XEN_GUEST_HANDLE_64
645 #define XEN_GUEST_HANDLE_64(name) XEN_GUEST_HANDLE(name)
646 #endif
647 #endif
649 #endif /* __XEN_PUBLIC_XEN_H__ */
651 /*
652 * Local variables:
653 * mode: C
654 * c-set-style: "BSD"
655 * c-basic-offset: 4
656 * tab-width: 4
657 * indent-tabs-mode: nil
658 * End:
659 */