1 /******************************************************************************
4 * Guest OS interface to Xen.
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:
13 * The above copyright notice and this permission notice shall be included in
14 * all copies or substantial portions of the Software.
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.
24 * Copyright (c) 2004, K A Fraser
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 #elif defined(__arm__)
39 #error "Unsupported architecture"
43 /* Guest handles for primitive C types. */
44 DEFINE_XEN_GUEST_HANDLE(char);
45 __DEFINE_XEN_GUEST_HANDLE(uchar, unsigned char);
46 DEFINE_XEN_GUEST_HANDLE(int);
47 __DEFINE_XEN_GUEST_HANDLE(uint, unsigned int);
48 DEFINE_XEN_GUEST_HANDLE(long);
49 __DEFINE_XEN_GUEST_HANDLE(ulong, unsigned long);
50 DEFINE_XEN_GUEST_HANDLE(void);
52 DEFINE_XEN_GUEST_HANDLE(uint64_t);
53 DEFINE_XEN_GUEST_HANDLE(xen_pfn_t);
60 /* `incontents 100 hcalls List of hypercalls
61 * ` enum hypercall_num { // __HYPERVISOR_* => HYPERVISOR_*()
64 #define __HYPERVISOR_set_trap_table 0
65 #define __HYPERVISOR_mmu_update 1
66 #define __HYPERVISOR_set_gdt 2
67 #define __HYPERVISOR_stack_switch 3
68 #define __HYPERVISOR_set_callbacks 4
69 #define __HYPERVISOR_fpu_taskswitch 5
70 #define __HYPERVISOR_sched_op_compat 6 /* compat since 0x00030101 */
71 #define __HYPERVISOR_platform_op 7
72 #define __HYPERVISOR_set_debugreg 8
73 #define __HYPERVISOR_get_debugreg 9
74 #define __HYPERVISOR_update_descriptor 10
75 #define __HYPERVISOR_memory_op 12
76 #define __HYPERVISOR_multicall 13
77 #define __HYPERVISOR_update_va_mapping 14
78 #define __HYPERVISOR_set_timer_op 15
79 #define __HYPERVISOR_event_channel_op_compat 16 /* compat since 0x00030202 */
80 #define __HYPERVISOR_xen_version 17
81 #define __HYPERVISOR_console_io 18
82 #define __HYPERVISOR_physdev_op_compat 19 /* compat since 0x00030202 */
83 #define __HYPERVISOR_grant_table_op 20
84 #define __HYPERVISOR_vm_assist 21
85 #define __HYPERVISOR_update_va_mapping_otherdomain 22
86 #define __HYPERVISOR_iret 23 /* x86 only */
87 #define __HYPERVISOR_vcpu_op 24
88 #define __HYPERVISOR_set_segment_base 25 /* x86/64 only */
89 #define __HYPERVISOR_mmuext_op 26
90 #define __HYPERVISOR_xsm_op 27
91 #define __HYPERVISOR_nmi_op 28
92 #define __HYPERVISOR_sched_op 29
93 #define __HYPERVISOR_callback_op 30
94 #define __HYPERVISOR_xenoprof_op 31
95 #define __HYPERVISOR_event_channel_op 32
96 #define __HYPERVISOR_physdev_op 33
97 #define __HYPERVISOR_hvm_op 34
98 #define __HYPERVISOR_sysctl 35
99 #define __HYPERVISOR_domctl 36
100 #define __HYPERVISOR_kexec_op 37
101 #define __HYPERVISOR_tmem_op 38
102 #define __HYPERVISOR_xc_reserved_op 39 /* reserved for XenClient */
104 /* Architecture-specific hypercall definitions. */
105 #define __HYPERVISOR_arch_0 48
106 #define __HYPERVISOR_arch_1 49
107 #define __HYPERVISOR_arch_2 50
108 #define __HYPERVISOR_arch_3 51
109 #define __HYPERVISOR_arch_4 52
110 #define __HYPERVISOR_arch_5 53
111 #define __HYPERVISOR_arch_6 54
112 #define __HYPERVISOR_arch_7 55
117 * HYPERCALL COMPATIBILITY.
120 /* New sched_op hypercall introduced in 0x00030101. */
121 #if __XEN_INTERFACE_VERSION__ < 0x00030101
122 #undef __HYPERVISOR_sched_op
123 #define __HYPERVISOR_sched_op __HYPERVISOR_sched_op_compat
126 /* New event-channel and physdev hypercalls introduced in 0x00030202. */
127 #if __XEN_INTERFACE_VERSION__ < 0x00030202
128 #undef __HYPERVISOR_event_channel_op
129 #define __HYPERVISOR_event_channel_op __HYPERVISOR_event_channel_op_compat
130 #undef __HYPERVISOR_physdev_op
131 #define __HYPERVISOR_physdev_op __HYPERVISOR_physdev_op_compat
134 /* New platform_op hypercall introduced in 0x00030204. */
135 #if __XEN_INTERFACE_VERSION__ < 0x00030204
136 #define __HYPERVISOR_dom0_op __HYPERVISOR_platform_op
142 * Virtual interrupts that a guest OS may receive from Xen.
144 * In the side comments, 'V.' denotes a per-VCPU VIRQ while 'G.' denotes a
145 * global VIRQ. The former can be bound once per VCPU and cannot be re-bound.
146 * The latter can be allocated only once per guest: they must initially be
147 * allocated to VCPU0 but can subsequently be re-bound.
150 #define VIRQ_TIMER 0 /* V. Timebase update, and/or requested timeout. */
151 #define VIRQ_DEBUG 1 /* V. Request guest to dump debug info. */
152 #define VIRQ_CONSOLE 2 /* G. (DOM0) Bytes received on emergency console. */
153 #define VIRQ_DOM_EXC 3 /* G. (DOM0) Exceptional event for some domain. */
154 #define VIRQ_TBUF 4 /* G. (DOM0) Trace buffer has records available. */
155 #define VIRQ_DEBUGGER 6 /* G. (DOM0) A domain has paused for debugging. */
156 #define VIRQ_XENOPROF 7 /* V. XenOprofile interrupt: new sample available */
157 #define VIRQ_CON_RING 8 /* G. (DOM0) Bytes received on console */
158 #define VIRQ_PCPU_STATE 9 /* G. (DOM0) PCPU state changed */
159 #define VIRQ_MEM_EVENT 10 /* G. (DOM0) A memory event has occured */
160 #define VIRQ_XC_RESERVED 11 /* G. Reserved for XenClient */
161 #define VIRQ_ENOMEM 12 /* G. (DOM0) Low on heap memory */
163 /* Architecture-specific VIRQ definitions. */
164 #define VIRQ_ARCH_0 16
165 #define VIRQ_ARCH_1 17
166 #define VIRQ_ARCH_2 18
167 #define VIRQ_ARCH_3 19
168 #define VIRQ_ARCH_4 20
169 #define VIRQ_ARCH_5 21
170 #define VIRQ_ARCH_6 22
171 #define VIRQ_ARCH_7 23
177 * ` enum neg_errnoval
178 * ` HYPERVISOR_mmu_update(const struct mmu_update reqs[],
179 * ` unsigned count, unsigned *done_out,
180 * ` unsigned foreigndom)
182 * @reqs is an array of mmu_update_t structures ((ptr, val) pairs).
183 * @count is the length of the above array.
184 * @pdone is an output parameter indicating number of completed operations
185 * @foreigndom[15:0]: FD, the expected owner of data pages referenced in this
186 * hypercall invocation. Can be DOMID_SELF.
187 * @foreigndom[31:16]: PFD, the expected owner of pagetable pages referenced
188 * in this hypercall invocation. The value of this field
189 * (x) encodes the PFD as follows:
190 * x == 0 => PFD == DOMID_SELF
191 * x != 0 => PFD == x - 1
193 * Sub-commands: ptr[1:0] specifies the appropriate MMU_* command.
195 * ptr[1:0] == MMU_NORMAL_PT_UPDATE:
196 * Updates an entry in a page table belonging to PFD. If updating an L1 table,
197 * and the new table entry is valid/present, the mapped frame must belong to
198 * FD. If attempting to map an I/O page then the caller assumes the privilege
200 * FD == DOMID_IO: Permit /only/ I/O mappings, at the priv level of the caller.
201 * FD == DOMID_XEN: Map restricted areas of Xen's heap space.
202 * ptr[:2] -- Machine address of the page-table entry to modify.
203 * val -- Value to write.
205 * There also certain implicit requirements when using this hypercall. The
206 * pages that make up a pagetable must be mapped read-only in the guest.
207 * This prevents uncontrolled guest updates to the pagetable. Xen strictly
208 * enforces this, and will disallow any pagetable update which will end up
209 * mapping pagetable page RW, and will disallow using any writable page as a
210 * pagetable. In practice it means that when constructing a page table for a
211 * process, thread, etc, we MUST be very dilligient in following these rules:
212 * 1). Start with top-level page (PGD or in Xen language: L4). Fill out
214 * 2). Keep on going, filling out the upper (PUD or L3), and middle (PMD
216 * 3). Start filling out the PTE table (L1) with the PTE entries. Once
217 * done, make sure to set each of those entries to RO (so writeable bit
218 * is unset). Once that has been completed, set the PMD (L2) for this
220 * 4). When completed with all of the PMD (L2) entries, and all of them have
221 * been set to RO, make sure to set RO the PUD (L3). Do the same
222 * operation on PGD (L4) pagetable entries that have a PUD (L3) entry.
223 * 5). Now before you can use those pages (so setting the cr3), you MUST also
224 * pin them so that the hypervisor can verify the entries. This is done
225 * via the HYPERVISOR_mmuext_op(MMUEXT_PIN_L4_TABLE, guest physical frame
226 * number of the PGD (L4)). And this point the HYPERVISOR_mmuext_op(
227 * MMUEXT_NEW_BASEPTR, guest physical frame number of the PGD (L4)) can be
229 * For 32-bit guests, the L4 is not used (as there is less pagetables), so
231 * At this point the pagetables can be modified using the MMU_NORMAL_PT_UPDATE
232 * hypercall. Also if so desired the OS can also try to write to the PTE
233 * and be trapped by the hypervisor (as the PTE entry is RO).
235 * To deallocate the pages, the operations are the reverse of the steps
236 * mentioned above. The argument is MMUEXT_UNPIN_TABLE for all levels and the
237 * pagetable MUST not be in use (meaning that the cr3 is not set to it).
239 * ptr[1:0] == MMU_MACHPHYS_UPDATE:
240 * Updates an entry in the machine->pseudo-physical mapping table.
241 * ptr[:2] -- Machine address within the frame whose mapping to modify.
242 * The frame must belong to the FD, if one is specified.
243 * val -- Value to write into the mapping entry.
245 * ptr[1:0] == MMU_PT_UPDATE_PRESERVE_AD:
246 * As MMU_NORMAL_PT_UPDATE above, but A/D bits currently in the PTE are ORed
247 * with those in @val.
249 * @val is usually the machine frame number along with some attributes.
250 * The attributes by default follow the architecture defined bits. Meaning that
251 * if this is a X86_64 machine and four page table layout is used, the layout
253 * - 63 if set means No execute (NX)
254 * - 46-13 the machine frame number
255 * - 12 available for guest
256 * - 11 available for guest
257 * - 10 available for guest
258 * - 9 available for guest
260 * - 7 PAT (PSE is disabled, must use hypercall to make 4MB or 2MB pages)
263 * - 4 page cached disabled
264 * - 3 page write through
265 * - 2 userspace accessible
269 * The one bits that does not fit with the default layout is the PAGE_PSE
270 * also called PAGE_PAT). The MMUEXT_[UN]MARK_SUPER arguments to the
271 * HYPERVISOR_mmuext_op serve as mechanism to set a pagetable to be 4MB
272 * (or 2MB) instead of using the PAGE_PSE bit.
274 * The reason that the PAGE_PSE (bit 7) is not being utilized is due to Xen
275 * using it as the Page Attribute Table (PAT) bit - for details on it please
276 * refer to Intel SDM 10.12. The PAT allows to set the caching attributes of
277 * pages instead of using MTRRs.
279 * The PAT MSR is as follow (it is a 64-bit value, each entry is 8 bits):
281 * +---+----+----+----+-----+----+----+
282 * WC | WC | WB | UC | UC- | WC | WB | <= Linux
283 * +---+----+----+----+-----+----+----+
284 * WC | WT | WB | UC | UC- | WT | WB | <= BIOS (default when machine boots)
285 * +---+----+----+----+-----+----+----+
286 * WC | WP | WC | UC | UC- | WT | WB | <= Xen
287 * +---+----+----+----+-----+----+----+
289 * The lookup of this index table translates to looking up
290 * Bit 7, Bit 4, and Bit 3 of val entry:
292 * PAT/PSE (bit 7) ... PCD (bit 4) .. PWT (bit 3).
294 * If all bits are off, then we are using PAT0. If bit 3 turned on,
295 * then we are using PAT1, if bit 3 and bit 4, then PAT2..
297 * As you can see, the Linux PAT1 translates to PAT4 under Xen. Which means
298 * that if a guest that follows Linux's PAT setup and would like to set Write
299 * Combined on pages it MUST use PAT4 entry. Meaning that Bit 7 (PAGE_PAT) is
300 * set. For example, under Linux it only uses PAT0, PAT1, and PAT2 for the
303 * WB = none (so PAT0)
304 * WC = PWT (bit 3 on)
305 * UC = PWT | PCD (bit 3 and 4 are on).
307 * To make it work with Xen, it needs to translate the WC bit as so:
309 * PWT (so bit 3 on) --> PAT (so bit 7 is on) and clear bit 3
311 * And to translate back it would:
313 * PAT (bit 7 on) --> PWT (bit 3 on) and clear bit 7.
315 #define MMU_NORMAL_PT_UPDATE 0 /* checked '*ptr = val'. ptr is MA. */
316 #define MMU_MACHPHYS_UPDATE 1 /* ptr = MA of frame to modify entry for */
317 #define MMU_PT_UPDATE_PRESERVE_AD 2 /* atomically: *ptr = val | (*ptr&(A|D)) */
320 * MMU EXTENDED OPERATIONS
322 * HYPERVISOR_mmuext_op() accepts a list of mmuext_op structures.
323 * A foreigndom (FD) can be specified (or DOMID_SELF for none).
324 * Where the FD has some effect, it is described below.
326 * cmd: MMUEXT_(UN)PIN_*_TABLE
327 * mfn: Machine frame number to be (un)pinned as a p.t. page.
328 * The frame must belong to the FD, if one is specified.
330 * cmd: MMUEXT_NEW_BASEPTR
331 * mfn: Machine frame number of new page-table base to install in MMU.
333 * cmd: MMUEXT_NEW_USER_BASEPTR [x86/64 only]
334 * mfn: Machine frame number of new page-table base to install in MMU
335 * when in user space.
337 * cmd: MMUEXT_TLB_FLUSH_LOCAL
338 * No additional arguments. Flushes local TLB.
340 * cmd: MMUEXT_INVLPG_LOCAL
341 * linear_addr: Linear address to be flushed from the local TLB.
343 * cmd: MMUEXT_TLB_FLUSH_MULTI
344 * vcpumask: Pointer to bitmap of VCPUs to be flushed.
346 * cmd: MMUEXT_INVLPG_MULTI
347 * linear_addr: Linear address to be flushed.
348 * vcpumask: Pointer to bitmap of VCPUs to be flushed.
350 * cmd: MMUEXT_TLB_FLUSH_ALL
351 * No additional arguments. Flushes all VCPUs' TLBs.
353 * cmd: MMUEXT_INVLPG_ALL
354 * linear_addr: Linear address to be flushed from all VCPUs' TLBs.
356 * cmd: MMUEXT_FLUSH_CACHE
357 * No additional arguments. Writes back and flushes cache contents.
359 * cmd: MMUEXT_FLUSH_CACHE_GLOBAL
360 * No additional arguments. Writes back and flushes cache contents
361 * on all CPUs in the system.
363 * cmd: MMUEXT_SET_LDT
364 * linear_addr: Linear address of LDT base (NB. must be page-aligned).
365 * nr_ents: Number of entries in LDT.
367 * cmd: MMUEXT_CLEAR_PAGE
368 * mfn: Machine frame number to be cleared.
370 * cmd: MMUEXT_COPY_PAGE
371 * mfn: Machine frame number of the destination page.
372 * src_mfn: Machine frame number of the source page.
374 * cmd: MMUEXT_[UN]MARK_SUPER
375 * mfn: Machine frame number of head of superpage to be [un]marked.
377 #define MMUEXT_PIN_L1_TABLE 0
378 #define MMUEXT_PIN_L2_TABLE 1
379 #define MMUEXT_PIN_L3_TABLE 2
380 #define MMUEXT_PIN_L4_TABLE 3
381 #define MMUEXT_UNPIN_TABLE 4
382 #define MMUEXT_NEW_BASEPTR 5
383 #define MMUEXT_TLB_FLUSH_LOCAL 6
384 #define MMUEXT_INVLPG_LOCAL 7
385 #define MMUEXT_TLB_FLUSH_MULTI 8
386 #define MMUEXT_INVLPG_MULTI 9
387 #define MMUEXT_TLB_FLUSH_ALL 10
388 #define MMUEXT_INVLPG_ALL 11
389 #define MMUEXT_FLUSH_CACHE 12
390 #define MMUEXT_SET_LDT 13
391 #define MMUEXT_NEW_USER_BASEPTR 15
392 #define MMUEXT_CLEAR_PAGE 16
393 #define MMUEXT_COPY_PAGE 17
394 #define MMUEXT_FLUSH_CACHE_GLOBAL 18
395 #define MMUEXT_MARK_SUPER 19
396 #define MMUEXT_UNMARK_SUPER 20
402 /* [UN]PIN_TABLE, NEW_BASEPTR, NEW_USER_BASEPTR
403 * CLEAR_PAGE, COPY_PAGE, [UN]MARK_SUPER */
405 /* INVLPG_LOCAL, INVLPG_ALL, SET_LDT */
406 unsigned long linear_addr;
410 unsigned int nr_ents;
411 /* TLB_FLUSH_MULTI, INVLPG_MULTI */
412 #if __XEN_INTERFACE_VERSION__ >= 0x00030205
413 XEN_GUEST_HANDLE(const_void) vcpumask;
415 const void *vcpumask;
421 typedef struct mmuext_op mmuext_op_t;
422 DEFINE_XEN_GUEST_HANDLE(mmuext_op_t);
425 /* These are passed as 'flags' to update_va_mapping. They can be ORed. */
426 /* When specifying UVMF_MULTI, also OR in a pointer to a CPU bitmap. */
427 /* UVMF_LOCAL is merely UVMF_MULTI with a NULL bitmap pointer. */
428 #define UVMF_NONE (0UL<<0) /* No flushing at all. */
429 #define UVMF_TLB_FLUSH (1UL<<0) /* Flush entire TLB(s). */
430 #define UVMF_INVLPG (2UL<<0) /* Flush only one entry. */
431 #define UVMF_FLUSHTYPE_MASK (3UL<<0)
432 #define UVMF_MULTI (0UL<<2) /* Flush subset of TLBs. */
433 #define UVMF_LOCAL (0UL<<2) /* Flush local TLB. */
434 #define UVMF_ALL (1UL<<2) /* Flush all TLBs. */
437 * Commands to HYPERVISOR_console_io().
439 #define CONSOLEIO_write 0
440 #define CONSOLEIO_read 1
443 * Commands to HYPERVISOR_vm_assist().
445 #define VMASST_CMD_enable 0
446 #define VMASST_CMD_disable 1
448 /* x86/32 guests: simulate full 4GB segment limits. */
449 #define VMASST_TYPE_4gb_segments 0
451 /* x86/32 guests: trap (vector 15) whenever above vmassist is used. */
452 #define VMASST_TYPE_4gb_segments_notify 1
455 * x86 guests: support writes to bottom-level PTEs.
456 * NB1. Page-directory entries cannot be written.
457 * NB2. Guest must continue to remove all writable mappings of PTEs.
459 #define VMASST_TYPE_writable_pagetables 2
461 /* x86/PAE guests: support PDPTs above 4GB. */
462 #define VMASST_TYPE_pae_extended_cr3 3
464 #define MAX_VMASST_TYPE 3
468 typedef uint16_t domid_t;
470 /* Domain ids >= DOMID_FIRST_RESERVED cannot be used for ordinary domains. */
471 #define DOMID_FIRST_RESERVED (0x7FF0U)
473 /* DOMID_SELF is used in certain contexts to refer to oneself. */
474 #define DOMID_SELF (0x7FF0U)
477 * DOMID_IO is used to restrict page-table updates to mapping I/O memory.
478 * Although no Foreign Domain need be specified to map I/O pages, DOMID_IO
479 * is useful to ensure that no mappings to the OS's own heap are accidentally
480 * installed. (e.g., in Linux this could cause havoc as reference counts
481 * aren't adjusted on the I/O-mapping code path).
482 * This only makes sense in MMUEXT_SET_FOREIGNDOM, but in that context can
483 * be specified by any calling domain.
485 #define DOMID_IO (0x7FF1U)
488 * DOMID_XEN is used to allow privileged domains to map restricted parts of
489 * Xen's heap space (e.g., the machine_to_phys table).
490 * This only makes sense in MMUEXT_SET_FOREIGNDOM, and is only permitted if
491 * the caller is privileged.
493 #define DOMID_XEN (0x7FF2U)
496 * DOMID_COW is used as the owner of sharable pages */
497 #define DOMID_COW (0x7FF3U)
499 /* DOMID_INVALID is used to identify pages with unknown owner. */
500 #define DOMID_INVALID (0x7FF4U)
503 #define DOMID_IDLE (0x7FFFU)
506 * Send an array of these to HYPERVISOR_mmu_update().
507 * NB. The fields are natural pointer/address size for this architecture.
510 uint64_t ptr; /* Machine address of PTE. */
511 uint64_t val; /* New contents of PTE. */
513 typedef struct mmu_update mmu_update_t;
514 DEFINE_XEN_GUEST_HANDLE(mmu_update_t);
517 * Send an array of these to HYPERVISOR_multicall().
518 * NB. The fields are natural register size for this architecture.
520 struct multicall_entry {
521 unsigned long op, result;
522 unsigned long args[6];
524 typedef struct multicall_entry multicall_entry_t;
525 DEFINE_XEN_GUEST_HANDLE(multicall_entry_t);
528 * Event channel endpoints per domain:
529 * 1024 if a long is 32 bits; 4096 if a long is 64 bits.
531 #define NR_EVENT_CHANNELS (sizeof(unsigned long) * sizeof(unsigned long) * 64)
533 struct vcpu_time_info {
535 * Updates to the following values are preceded and followed by an
536 * increment of 'version'. The guest can therefore detect updates by
537 * looking for changes to 'version'. If the least-significant bit of
538 * the version number is set then an update is in progress and the guest
539 * must wait to read a consistent set of values.
540 * The correct way to interact with the version number is similar to
541 * Linux's seqlock: see the implementations of read_seqbegin/read_seqretry.
545 uint64_t tsc_timestamp; /* TSC at last update of time vals. */
546 uint64_t system_time; /* Time, in nanosecs, since boot. */
548 * Current system time:
550 * ((((tsc - tsc_timestamp) << tsc_shift) * tsc_to_system_mul) >> 32)
551 * CPU frequency (Hz):
552 * ((10^9 << 32) / tsc_to_system_mul) >> tsc_shift
554 uint32_t tsc_to_system_mul;
558 typedef struct vcpu_time_info vcpu_time_info_t;
562 * 'evtchn_upcall_pending' is written non-zero by Xen to indicate
563 * a pending notification for a particular VCPU. It is then cleared
564 * by the guest OS /before/ checking for pending work, thus avoiding
565 * a set-and-check race. Note that the mask is only accessed by Xen
566 * on the CPU that is currently hosting the VCPU. This means that the
567 * pending and mask flags can be updated by the guest without special
568 * synchronisation (i.e., no need for the x86 LOCK prefix).
569 * This may seem suboptimal because if the pending flag is set by
570 * a different CPU then an IPI may be scheduled even when the mask
571 * is set. However, note:
572 * 1. The task of 'interrupt holdoff' is covered by the per-event-
573 * channel mask bits. A 'noisy' event that is continually being
574 * triggered can be masked at source at this very precise
576 * 2. The main purpose of the per-VCPU mask is therefore to restrict
577 * reentrant execution: whether for concurrency control, or to
578 * prevent unbounded stack usage. Whatever the purpose, we expect
579 * that the mask will be asserted only for short periods at a time,
580 * and so the likelihood of a 'spurious' IPI is suitably small.
581 * The mask is read before making an event upcall to the guest: a
582 * non-zero mask therefore guarantees that the VCPU will not receive
583 * an upcall activation. The mask is cleared when the VCPU requests
584 * to block: this avoids wakeup-waiting races.
586 uint8_t evtchn_upcall_pending;
587 uint8_t evtchn_upcall_mask;
588 unsigned long evtchn_pending_sel;
589 struct arch_vcpu_info arch;
590 struct vcpu_time_info time;
591 }; /* 64 bytes (x86) */
593 typedef struct vcpu_info vcpu_info_t;
597 * Xen/kernel shared data -- pointer provided in start_info.
599 * This structure is defined to be both smaller than a page, and the
600 * only data on the shared page, but may vary in actual size even within
601 * compatible Xen versions; guests should not rely on the size
602 * of this structure remaining constant.
605 struct vcpu_info vcpu_info[XEN_LEGACY_MAX_VCPUS];
608 * A domain can create "event channels" on which it can send and receive
609 * asynchronous event notifications. There are three classes of event that
610 * are delivered by this mechanism:
611 * 1. Bi-directional inter- and intra-domain connections. Domains must
612 * arrange out-of-band to set up a connection (usually by allocating
613 * an unbound 'listener' port and avertising that via a storage service
615 * 2. Physical interrupts. A domain with suitable hardware-access
616 * privileges can bind an event-channel port to a physical interrupt
618 * 3. Virtual interrupts ('events'). A domain can bind an event-channel
619 * port to a virtual interrupt source, such as the virtual-timer
620 * device or the emergency console.
622 * Event channels are addressed by a "port index". Each channel is
623 * associated with two bits of information:
624 * 1. PENDING -- notifies the domain that there is a pending notification
625 * to be processed. This bit is cleared by the guest.
626 * 2. MASK -- if this bit is clear then a 0->1 transition of PENDING
627 * will cause an asynchronous upcall to be scheduled. This bit is only
628 * updated by the guest. It is read-only within Xen. If a channel
629 * becomes pending while the channel is masked then the 'edge' is lost
630 * (i.e., when the channel is unmasked, the guest must manually handle
631 * pending notifications as no upcall will be scheduled by Xen).
633 * To expedite scanning of pending notifications, any 0->1 pending
634 * transition on an unmasked channel causes a corresponding bit in a
635 * per-vcpu selector word to be set. Each bit in the selector covers a
636 * 'C long' in the PENDING bitfield array.
638 unsigned long evtchn_pending[sizeof(unsigned long) * 8];
639 unsigned long evtchn_mask[sizeof(unsigned long) * 8];
642 * Wallclock time: updated only by control software. Guests should base
643 * their gettimeofday() syscall on this wallclock-base value.
645 uint32_t wc_version; /* Version counter: see vcpu_time_info_t. */
646 uint32_t wc_sec; /* Secs 00:00:00 UTC, Jan 1, 1970. */
647 uint32_t wc_nsec; /* Nsecs 00:00:00 UTC, Jan 1, 1970. */
649 struct arch_shared_info arch;
653 typedef struct shared_info shared_info_t;
657 * Start-of-day memory layout:
658 * 1. The domain is started within contiguous virtual-memory region.
659 * 2. The contiguous region ends on an aligned 4MB boundary.
660 * 3. This the order of bootstrap elements in the initial virtual region:
661 * a. relocated kernel image
662 * b. initial ram disk [mod_start, mod_len]
663 * c. list of allocated page frames [mfn_list, nr_pages]
664 * (unless relocated due to XEN_ELFNOTE_INIT_P2M)
665 * d. start_info_t structure [register ESI (x86)]
666 * e. bootstrap page tables [pt_base, CR3 (x86)]
667 * f. bootstrap stack [register ESP (x86)]
668 * 4. Bootstrap elements are packed together, but each is 4kB-aligned.
669 * 5. The initial ram disk may be omitted.
670 * 6. The list of page frames forms a contiguous 'pseudo-physical' memory
671 * layout for the domain. In particular, the bootstrap virtual-memory
672 * region is a 1:1 mapping to the first section of the pseudo-physical map.
673 * 7. All bootstrap elements are mapped read-writable for the guest OS. The
674 * only exception is the bootstrap page table, which is mapped read-only.
675 * 8. There is guaranteed to be at least 512kB padding after the final
676 * bootstrap element. If necessary, the bootstrap virtual region is
677 * extended by an extra 4MB to ensure this.
680 #define MAX_GUEST_CMDLINE 1024
682 /* THE FOLLOWING ARE FILLED IN BOTH ON INITIAL BOOT AND ON RESUME. */
683 char magic[32]; /* "xen-<version>-<platform>". */
684 unsigned long nr_pages; /* Total pages allocated to this domain. */
685 unsigned long shared_info; /* MACHINE address of shared info struct. */
686 uint32_t flags; /* SIF_xxx flags. */
687 xen_pfn_t store_mfn; /* MACHINE page number of shared page. */
688 uint32_t store_evtchn; /* Event channel for store communication. */
691 xen_pfn_t mfn; /* MACHINE page number of console page. */
692 uint32_t evtchn; /* Event channel for console page. */
695 uint32_t info_off; /* Offset of console_info struct. */
696 uint32_t info_size; /* Size of console_info struct from start.*/
699 /* THE FOLLOWING ARE ONLY FILLED IN ON INITIAL BOOT (NOT RESUME). */
700 unsigned long pt_base; /* VIRTUAL address of page directory. */
701 unsigned long nr_pt_frames; /* Number of bootstrap p.t. frames. */
702 unsigned long mfn_list; /* VIRTUAL address of page-frame list. */
703 unsigned long mod_start; /* VIRTUAL address of pre-loaded module */
704 /* (PFN of pre-loaded module if */
705 /* SIF_MOD_START_PFN set in flags). */
706 unsigned long mod_len; /* Size (bytes) of pre-loaded module. */
707 int8_t cmd_line[MAX_GUEST_CMDLINE];
708 /* The pfn range here covers both page table and p->m table frames. */
709 unsigned long first_p2m_pfn;/* 1st pfn forming initial P->M table. */
710 unsigned long nr_p2m_frames;/* # of pfns forming initial P->M table. */
712 typedef struct start_info start_info_t;
714 /* New console union for dom0 introduced in 0x00030203. */
715 #if __XEN_INTERFACE_VERSION__ < 0x00030203
716 #define console_mfn console.domU.mfn
717 #define console_evtchn console.domU.evtchn
720 /* These flags are passed in the 'flags' field of start_info_t. */
721 #define SIF_PRIVILEGED (1<<0) /* Is the domain privileged? */
722 #define SIF_INITDOMAIN (1<<1) /* Is this the initial control domain? */
723 #define SIF_MULTIBOOT_MOD (1<<2) /* Is mod_start a multiboot module? */
724 #define SIF_MOD_START_PFN (1<<3) /* Is mod_start a PFN? */
725 #define SIF_PM_MASK (0xFF<<8) /* reserve 1 byte for xen-pm options */
728 * A multiboot module is a package containing modules very similar to a
729 * multiboot module array. The only differences are:
730 * - the array of module descriptors is by convention simply at the beginning
731 * of the multiboot module,
732 * - addresses in the module descriptors are based on the beginning of the
734 * - the number of modules is determined by a termination descriptor that has
737 * This permits to both build it statically and reference it in a configuration
738 * file, and let the PV guest easily rebase the addresses to virtual addresses
739 * and at the same time count the number of modules.
741 struct xen_multiboot_mod_list
743 /* Address of first byte of the module */
745 /* Address of last byte of the module (inclusive) */
747 /* Address of zero-terminated command line */
749 /* Unused, must be zero */
753 typedef struct dom0_vga_console_info {
754 uint8_t video_type; /* DOM0_VGA_CONSOLE_??? */
755 #define XEN_VGATYPE_TEXT_MODE_3 0x03
756 #define XEN_VGATYPE_VESA_LFB 0x23
757 #define XEN_VGATYPE_EFI_LFB 0x70
761 /* Font height, in pixels. */
762 uint16_t font_height;
763 /* Cursor location (column, row). */
764 uint16_t cursor_x, cursor_y;
765 /* Number of rows and columns (dimensions in characters). */
766 uint16_t rows, columns;
770 /* Width and height, in pixels. */
771 uint16_t width, height;
772 /* Bytes per scan line. */
773 uint16_t bytes_per_line;
774 /* Bits per pixel. */
775 uint16_t bits_per_pixel;
776 /* LFB physical address, and size (in units of 64kB). */
779 /* RGB mask offsets and sizes, as defined by VBE 1.2+ */
780 uint8_t red_pos, red_size;
781 uint8_t green_pos, green_size;
782 uint8_t blue_pos, blue_size;
783 uint8_t rsvd_pos, rsvd_size;
784 #if __XEN_INTERFACE_VERSION__ >= 0x00030206
785 /* VESA capabilities (offset 0xa, VESA command 0x4f00). */
787 /* Mode attributes (offset 0x0, VESA command 0x4f01). */
792 } dom0_vga_console_info_t;
793 #define xen_vga_console_info dom0_vga_console_info
794 #define xen_vga_console_info_t dom0_vga_console_info_t
796 typedef uint8_t xen_domain_handle_t[16];
798 /* Turn a plain number into a C unsigned long constant. */
799 #define __mk_unsigned_long(x) x ## UL
800 #define mk_unsigned_long(x) __mk_unsigned_long(x)
802 __DEFINE_XEN_GUEST_HANDLE(uint8, uint8_t);
803 __DEFINE_XEN_GUEST_HANDLE(uint16, uint16_t);
804 __DEFINE_XEN_GUEST_HANDLE(uint32, uint32_t);
805 __DEFINE_XEN_GUEST_HANDLE(uint64, uint64_t);
807 #else /* __ASSEMBLY__ */
809 /* In assembly code we cannot use C numeric constant suffixes. */
810 #define mk_unsigned_long(x) x
812 #endif /* !__ASSEMBLY__ */
814 /* Default definitions for macros used by domctl/sysctl. */
815 #if defined(__XEN__) || defined(__XEN_TOOLS__)
817 #ifndef uint64_aligned_t
818 #define uint64_aligned_t uint64_t
820 #ifndef XEN_GUEST_HANDLE_64
821 #define XEN_GUEST_HANDLE_64(name) XEN_GUEST_HANDLE(name)
825 struct xenctl_cpumap {
826 XEN_GUEST_HANDLE_64(uint8) bitmap;
831 #endif /* defined(__XEN__) || defined(__XEN_TOOLS__) */
833 #endif /* __XEN_PUBLIC_XEN_H__ */
841 * indent-tabs-mode: nil