![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to xen.h CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [GnuMach] / Gnu-Mach / xen / public|
Return to xen.h CVS log | Up to [GnuMach] / Gnu-Mach / xen / public |
1.1 root 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: */
26:
27: #ifndef __XEN_PUBLIC_XEN_H__
28: #define __XEN_PUBLIC_XEN_H__
29:
30: #include <sys/types.h>
31:
32: #include "xen-compat.h"
33:
34: #if defined(__i386__) || defined(__x86_64__)
35: #include "arch-x86/xen.h"
36: #elif defined(__ia64__)
37: #include "arch-ia64.h"
38: #else
39: #error "Unsupported architecture"
40: #endif
41:
42: #ifndef __ASSEMBLY__
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);
51:
52: DEFINE_XEN_GUEST_HANDLE(xen_pfn_t);
53: #endif
54:
55: /*
56: * HYPERCALLS
57: */
58:
59: #define __HYPERVISOR_set_trap_table 0
60: #define __HYPERVISOR_mmu_update 1
61: #define __HYPERVISOR_set_gdt 2
62: #define __HYPERVISOR_stack_switch 3
63: #define __HYPERVISOR_set_callbacks 4
64: #define __HYPERVISOR_fpu_taskswitch 5
65: #define __HYPERVISOR_sched_op_compat 6 /* compat since 0x00030101 */
66: #define __HYPERVISOR_platform_op 7
67: #define __HYPERVISOR_set_debugreg 8
68: #define __HYPERVISOR_get_debugreg 9
69: #define __HYPERVISOR_update_descriptor 10
70: #define __HYPERVISOR_memory_op 12
71: #define __HYPERVISOR_multicall 13
72: #define __HYPERVISOR_update_va_mapping 14
73: #define __HYPERVISOR_set_timer_op 15
74: #define __HYPERVISOR_event_channel_op_compat 16 /* compat since 0x00030202 */
75: #define __HYPERVISOR_xen_version 17
76: #define __HYPERVISOR_console_io 18
77: #define __HYPERVISOR_physdev_op_compat 19 /* compat since 0x00030202 */
78: #define __HYPERVISOR_grant_table_op 20
79: #define __HYPERVISOR_vm_assist 21
80: #define __HYPERVISOR_update_va_mapping_otherdomain 22
81: #define __HYPERVISOR_iret 23 /* x86 only */
82: #define __HYPERVISOR_vcpu_op 24
83: #define __HYPERVISOR_set_segment_base 25 /* x86/64 only */
84: #define __HYPERVISOR_mmuext_op 26
85: #define __HYPERVISOR_xsm_op 27
86: #define __HYPERVISOR_nmi_op 28
87: #define __HYPERVISOR_sched_op 29
88: #define __HYPERVISOR_callback_op 30
89: #define __HYPERVISOR_xenoprof_op 31
90: #define __HYPERVISOR_event_channel_op 32
91: #define __HYPERVISOR_physdev_op 33
92: #define __HYPERVISOR_hvm_op 34
93: #define __HYPERVISOR_sysctl 35
94: #define __HYPERVISOR_domctl 36
95: #define __HYPERVISOR_kexec_op 37
96:
97: /* Architecture-specific hypercall definitions. */
98: #define __HYPERVISOR_arch_0 48
99: #define __HYPERVISOR_arch_1 49
100: #define __HYPERVISOR_arch_2 50
101: #define __HYPERVISOR_arch_3 51
102: #define __HYPERVISOR_arch_4 52
103: #define __HYPERVISOR_arch_5 53
104: #define __HYPERVISOR_arch_6 54
105: #define __HYPERVISOR_arch_7 55
106:
107: /*
108: * HYPERCALL COMPATIBILITY.
109: */
110:
111: /* New sched_op hypercall introduced in 0x00030101. */
112: #if __XEN_INTERFACE_VERSION__ < 0x00030101
113: #undef __HYPERVISOR_sched_op
114: #define __HYPERVISOR_sched_op __HYPERVISOR_sched_op_compat
115: #endif
116:
117: /* New event-channel and physdev hypercalls introduced in 0x00030202. */
118: #if __XEN_INTERFACE_VERSION__ < 0x00030202
119: #undef __HYPERVISOR_event_channel_op
120: #define __HYPERVISOR_event_channel_op __HYPERVISOR_event_channel_op_compat
121: #undef __HYPERVISOR_physdev_op
122: #define __HYPERVISOR_physdev_op __HYPERVISOR_physdev_op_compat
123: #endif
124:
125: /* New platform_op hypercall introduced in 0x00030204. */
126: #if __XEN_INTERFACE_VERSION__ < 0x00030204
127: #define __HYPERVISOR_dom0_op __HYPERVISOR_platform_op
128: #endif
129:
130: /*
131: * VIRTUAL INTERRUPTS
132: *
133: * Virtual interrupts that a guest OS may receive from Xen.
134: *
135: * In the side comments, 'V.' denotes a per-VCPU VIRQ while 'G.' denotes a
136: * global VIRQ. The former can be bound once per VCPU and cannot be re-bound.
137: * The latter can be allocated only once per guest: they must initially be
138: * allocated to VCPU0 but can subsequently be re-bound.
139: */
140: #define VIRQ_TIMER 0 /* V. Timebase update, and/or requested timeout. */
141: #define VIRQ_DEBUG 1 /* V. Request guest to dump debug info. */
142: #define VIRQ_CONSOLE 2 /* G. (DOM0) Bytes received on emergency console. */
143: #define VIRQ_DOM_EXC 3 /* G. (DOM0) Exceptional event for some domain. */
144: #define VIRQ_TBUF 4 /* G. (DOM0) Trace buffer has records available. */
145: #define VIRQ_DEBUGGER 6 /* G. (DOM0) A domain has paused for debugging. */
146: #define VIRQ_XENOPROF 7 /* V. XenOprofile interrupt: new sample available */
147: #define VIRQ_CON_RING 8 /* G. (DOM0) Bytes received on console */
148:
149: /* Architecture-specific VIRQ definitions. */
150: #define VIRQ_ARCH_0 16
151: #define VIRQ_ARCH_1 17
152: #define VIRQ_ARCH_2 18
153: #define VIRQ_ARCH_3 19
154: #define VIRQ_ARCH_4 20
155: #define VIRQ_ARCH_5 21
156: #define VIRQ_ARCH_6 22
157: #define VIRQ_ARCH_7 23
158:
159: #define NR_VIRQS 24
160:
161: /*
162: * MMU-UPDATE REQUESTS
163: *
164: * HYPERVISOR_mmu_update() accepts a list of (ptr, val) pairs.
165: * A foreigndom (FD) can be specified (or DOMID_SELF for none).
166: * Where the FD has some effect, it is described below.
167: * ptr[1:0] specifies the appropriate MMU_* command.
168: *
169: * ptr[1:0] == MMU_NORMAL_PT_UPDATE:
170: * Updates an entry in a page table. If updating an L1 table, and the new
171: * table entry is valid/present, the mapped frame must belong to the FD, if
172: * an FD has been specified. If attempting to map an I/O page then the
173: * caller assumes the privilege of the FD.
174: * FD == DOMID_IO: Permit /only/ I/O mappings, at the priv level of the caller.
175: * FD == DOMID_XEN: Map restricted areas of Xen's heap space.
176: * ptr[:2] -- Machine address of the page-table entry to modify.
177: * val -- Value to write.
178: *
179: * ptr[1:0] == MMU_MACHPHYS_UPDATE:
180: * Updates an entry in the machine->pseudo-physical mapping table.
181: * ptr[:2] -- Machine address within the frame whose mapping to modify.
182: * The frame must belong to the FD, if one is specified.
183: * val -- Value to write into the mapping entry.
184: *
185: * ptr[1:0] == MMU_PT_UPDATE_PRESERVE_AD:
186: * As MMU_NORMAL_PT_UPDATE above, but A/D bits currently in the PTE are ORed
187: * with those in @val.
188: */
189: #define MMU_NORMAL_PT_UPDATE 0 /* checked '*ptr = val'. ptr is MA. */
190: #define MMU_MACHPHYS_UPDATE 1 /* ptr = MA of frame to modify entry for */
191: #define MMU_PT_UPDATE_PRESERVE_AD 2 /* atomically: *ptr = val | (*ptr&(A|D)) */
192:
193: /*
194: * MMU EXTENDED OPERATIONS
195: *
196: * HYPERVISOR_mmuext_op() accepts a list of mmuext_op structures.
197: * A foreigndom (FD) can be specified (or DOMID_SELF for none).
198: * Where the FD has some effect, it is described below.
199: *
200: * cmd: MMUEXT_(UN)PIN_*_TABLE
201: * mfn: Machine frame number to be (un)pinned as a p.t. page.
202: * The frame must belong to the FD, if one is specified.
203: *
204: * cmd: MMUEXT_NEW_BASEPTR
205: * mfn: Machine frame number of new page-table base to install in MMU.
206: *
207: * cmd: MMUEXT_NEW_USER_BASEPTR [x86/64 only]
208: * mfn: Machine frame number of new page-table base to install in MMU
209: * when in user space.
210: *
211: * cmd: MMUEXT_TLB_FLUSH_LOCAL
212: * No additional arguments. Flushes local TLB.
213: *
214: * cmd: MMUEXT_INVLPG_LOCAL
215: * linear_addr: Linear address to be flushed from the local TLB.
216: *
217: * cmd: MMUEXT_TLB_FLUSH_MULTI
218: * vcpumask: Pointer to bitmap of VCPUs to be flushed.
219: *
220: * cmd: MMUEXT_INVLPG_MULTI
221: * linear_addr: Linear address to be flushed.
222: * vcpumask: Pointer to bitmap of VCPUs to be flushed.
223: *
224: * cmd: MMUEXT_TLB_FLUSH_ALL
225: * No additional arguments. Flushes all VCPUs' TLBs.
226: *
227: * cmd: MMUEXT_INVLPG_ALL
228: * linear_addr: Linear address to be flushed from all VCPUs' TLBs.
229: *
230: * cmd: MMUEXT_FLUSH_CACHE
231: * No additional arguments. Writes back and flushes cache contents.
232: *
233: * cmd: MMUEXT_SET_LDT
234: * linear_addr: Linear address of LDT base (NB. must be page-aligned).
235: * nr_ents: Number of entries in LDT.
236: *
237: * cmd: MMUEXT_CLEAR_PAGE
238: * mfn: Machine frame number to be cleared.
239: *
240: * cmd: MMUEXT_COPY_PAGE
241: * mfn: Machine frame number of the destination page.
242: * src_mfn: Machine frame number of the source page.
243: */
244: #define MMUEXT_PIN_L1_TABLE 0
245: #define MMUEXT_PIN_L2_TABLE 1
246: #define MMUEXT_PIN_L3_TABLE 2
247: #define MMUEXT_PIN_L4_TABLE 3
248: #define MMUEXT_UNPIN_TABLE 4
249: #define MMUEXT_NEW_BASEPTR 5
250: #define MMUEXT_TLB_FLUSH_LOCAL 6
251: #define MMUEXT_INVLPG_LOCAL 7
252: #define MMUEXT_TLB_FLUSH_MULTI 8
253: #define MMUEXT_INVLPG_MULTI 9
254: #define MMUEXT_TLB_FLUSH_ALL 10
255: #define MMUEXT_INVLPG_ALL 11
256: #define MMUEXT_FLUSH_CACHE 12
257: #define MMUEXT_SET_LDT 13
258: #define MMUEXT_NEW_USER_BASEPTR 15
259: #define MMUEXT_CLEAR_PAGE 16
260: #define MMUEXT_COPY_PAGE 17
261:
262: #ifndef __ASSEMBLY__
263: struct mmuext_op {
264: unsigned int cmd;
265: union {
266: /* [UN]PIN_TABLE, NEW_BASEPTR, NEW_USER_BASEPTR
267: * CLEAR_PAGE, COPY_PAGE */
268: xen_pfn_t mfn;
269: /* INVLPG_LOCAL, INVLPG_ALL, SET_LDT */
270: unsigned long linear_addr;
271: } arg1;
272: union {
273: /* SET_LDT */
274: unsigned int nr_ents;
275: /* TLB_FLUSH_MULTI, INVLPG_MULTI */
276: #if __XEN_INTERFACE_VERSION__ >= 0x00030205
277: XEN_GUEST_HANDLE(void) vcpumask;
278: #else
279: void *vcpumask;
280: #endif
281: /* COPY_PAGE */
282: xen_pfn_t src_mfn;
283: } arg2;
284: };
285: typedef struct mmuext_op mmuext_op_t;
286: DEFINE_XEN_GUEST_HANDLE(mmuext_op_t);
287: #endif
288:
289: /* These are passed as 'flags' to update_va_mapping. They can be ORed. */
290: /* When specifying UVMF_MULTI, also OR in a pointer to a CPU bitmap. */
291: /* UVMF_LOCAL is merely UVMF_MULTI with a NULL bitmap pointer. */
292: #define UVMF_NONE (0UL<<0) /* No flushing at all. */
293: #define UVMF_TLB_FLUSH (1UL<<0) /* Flush entire TLB(s). */
294: #define UVMF_INVLPG (2UL<<0) /* Flush only one entry. */
295: #define UVMF_FLUSHTYPE_MASK (3UL<<0)
296: #define UVMF_MULTI (0UL<<2) /* Flush subset of TLBs. */
297: #define UVMF_LOCAL (0UL<<2) /* Flush local TLB. */
298: #define UVMF_ALL (1UL<<2) /* Flush all TLBs. */
299:
300: /*
301: * Commands to HYPERVISOR_console_io().
302: */
303: #define CONSOLEIO_write 0
304: #define CONSOLEIO_read 1
305:
306: /*
307: * Commands to HYPERVISOR_vm_assist().
308: */
309: #define VMASST_CMD_enable 0
310: #define VMASST_CMD_disable 1
311:
312: /* x86/32 guests: simulate full 4GB segment limits. */
313: #define VMASST_TYPE_4gb_segments 0
314:
315: /* x86/32 guests: trap (vector 15) whenever above vmassist is used. */
316: #define VMASST_TYPE_4gb_segments_notify 1
317:
318: /*
319: * x86 guests: support writes to bottom-level PTEs.
320: * NB1. Page-directory entries cannot be written.
321: * NB2. Guest must continue to remove all writable mappings of PTEs.
322: */
323: #define VMASST_TYPE_writable_pagetables 2
324:
325: /* x86/PAE guests: support PDPTs above 4GB. */
326: #define VMASST_TYPE_pae_extended_cr3 3
327:
328: #define MAX_VMASST_TYPE 3
329:
330: #ifndef __ASSEMBLY__
331:
332: typedef uint16_t domid_t;
333:
334: /* Domain ids >= DOMID_FIRST_RESERVED cannot be used for ordinary domains. */
335: #define DOMID_FIRST_RESERVED (0x7FF0U)
336:
337: /* DOMID_SELF is used in certain contexts to refer to oneself. */
338: #define DOMID_SELF (0x7FF0U)
339:
340: /*
341: * DOMID_IO is used to restrict page-table updates to mapping I/O memory.
342: * Although no Foreign Domain need be specified to map I/O pages, DOMID_IO
343: * is useful to ensure that no mappings to the OS's own heap are accidentally
344: * installed. (e.g., in Linux this could cause havoc as reference counts
345: * aren't adjusted on the I/O-mapping code path).
346: * This only makes sense in MMUEXT_SET_FOREIGNDOM, but in that context can
347: * be specified by any calling domain.
348: */
349: #define DOMID_IO (0x7FF1U)
350:
351: /*
352: * DOMID_XEN is used to allow privileged domains to map restricted parts of
353: * Xen's heap space (e.g., the machine_to_phys table).
354: * This only makes sense in MMUEXT_SET_FOREIGNDOM, and is only permitted if
355: * the caller is privileged.
356: */
357: #define DOMID_XEN (0x7FF2U)
358:
359: /*
360: * Send an array of these to HYPERVISOR_mmu_update().
361: * NB. The fields are natural pointer/address size for this architecture.
362: */
363: struct mmu_update {
364: uint64_t ptr; /* Machine address of PTE. */
365: uint64_t val; /* New contents of PTE. */
366: };
367: typedef struct mmu_update mmu_update_t;
368: DEFINE_XEN_GUEST_HANDLE(mmu_update_t);
369:
370: /*
371: * Send an array of these to HYPERVISOR_multicall().
372: * NB. The fields are natural register size for this architecture.
373: */
374: struct multicall_entry {
375: unsigned long op, result;
376: unsigned long args[6];
377: };
378: typedef struct multicall_entry multicall_entry_t;
379: DEFINE_XEN_GUEST_HANDLE(multicall_entry_t);
380:
381: /*
382: * Event channel endpoints per domain:
383: * 1024 if a long is 32 bits; 4096 if a long is 64 bits.
384: */
385: #define NR_EVENT_CHANNELS (sizeof(unsigned long) * sizeof(unsigned long) * 64)
386:
387: struct vcpu_time_info {
388: /*
389: * Updates to the following values are preceded and followed by an
390: * increment of 'version'. The guest can therefore detect updates by
391: * looking for changes to 'version'. If the least-significant bit of
392: * the version number is set then an update is in progress and the guest
393: * must wait to read a consistent set of values.
394: * The correct way to interact with the version number is similar to
395: * Linux's seqlock: see the implementations of read_seqbegin/read_seqretry.
396: */
397: uint32_t version;
398: uint32_t pad0;
399: uint64_t tsc_timestamp; /* TSC at last update of time vals. */
400: uint64_t system_time; /* Time, in nanosecs, since boot. */
401: /*
402: * Current system time:
403: * system_time +
404: * ((((tsc - tsc_timestamp) << tsc_shift) * tsc_to_system_mul) >> 32)
405: * CPU frequency (Hz):
406: * ((10^9 << 32) / tsc_to_system_mul) >> tsc_shift
407: */
408: uint32_t tsc_to_system_mul;
409: int8_t tsc_shift;
410: int8_t pad1[3];
411: }; /* 32 bytes */
412: typedef struct vcpu_time_info vcpu_time_info_t;
413:
414: struct vcpu_info {
415: /*
416: * 'evtchn_upcall_pending' is written non-zero by Xen to indicate
417: * a pending notification for a particular VCPU. It is then cleared
418: * by the guest OS /before/ checking for pending work, thus avoiding
419: * a set-and-check race. Note that the mask is only accessed by Xen
420: * on the CPU that is currently hosting the VCPU. This means that the
421: * pending and mask flags can be updated by the guest without special
422: * synchronisation (i.e., no need for the x86 LOCK prefix).
423: * This may seem suboptimal because if the pending flag is set by
424: * a different CPU then an IPI may be scheduled even when the mask
425: * is set. However, note:
426: * 1. The task of 'interrupt holdoff' is covered by the per-event-
427: * channel mask bits. A 'noisy' event that is continually being
428: * triggered can be masked at source at this very precise
429: * granularity.
430: * 2. The main purpose of the per-VCPU mask is therefore to restrict
431: * reentrant execution: whether for concurrency control, or to
432: * prevent unbounded stack usage. Whatever the purpose, we expect
433: * that the mask will be asserted only for short periods at a time,
434: * and so the likelihood of a 'spurious' IPI is suitably small.
435: * The mask is read before making an event upcall to the guest: a
436: * non-zero mask therefore guarantees that the VCPU will not receive
437: * an upcall activation. The mask is cleared when the VCPU requests
438: * to block: this avoids wakeup-waiting races.
439: */
440: uint8_t evtchn_upcall_pending;
441: uint8_t evtchn_upcall_mask;
442: unsigned long evtchn_pending_sel;
443: struct arch_vcpu_info arch;
444: struct vcpu_time_info time;
445: }; /* 64 bytes (x86) */
446: #ifndef __XEN__
447: typedef struct vcpu_info vcpu_info_t;
448: #endif
449:
450: /*
451: * Xen/kernel shared data -- pointer provided in start_info.
452: *
453: * This structure is defined to be both smaller than a page, and the
454: * only data on the shared page, but may vary in actual size even within
455: * compatible Xen versions; guests should not rely on the size
456: * of this structure remaining constant.
457: */
458: struct shared_info {
459: struct vcpu_info vcpu_info[MAX_VIRT_CPUS];
460:
461: /*
462: * A domain can create "event channels" on which it can send and receive
463: * asynchronous event notifications. There are three classes of event that
464: * are delivered by this mechanism:
465: * 1. Bi-directional inter- and intra-domain connections. Domains must
466: * arrange out-of-band to set up a connection (usually by allocating
467: * an unbound 'listener' port and avertising that via a storage service
468: * such as xenstore).
469: * 2. Physical interrupts. A domain with suitable hardware-access
470: * privileges can bind an event-channel port to a physical interrupt
471: * source.
472: * 3. Virtual interrupts ('events'). A domain can bind an event-channel
473: * port to a virtual interrupt source, such as the virtual-timer
474: * device or the emergency console.
475: *
476: * Event channels are addressed by a "port index". Each channel is
477: * associated with two bits of information:
478: * 1. PENDING -- notifies the domain that there is a pending notification
479: * to be processed. This bit is cleared by the guest.
480: * 2. MASK -- if this bit is clear then a 0->1 transition of PENDING
481: * will cause an asynchronous upcall to be scheduled. This bit is only
482: * updated by the guest. It is read-only within Xen. If a channel
483: * becomes pending while the channel is masked then the 'edge' is lost
484: * (i.e., when the channel is unmasked, the guest must manually handle
485: * pending notifications as no upcall will be scheduled by Xen).
486: *
487: * To expedite scanning of pending notifications, any 0->1 pending
488: * transition on an unmasked channel causes a corresponding bit in a
489: * per-vcpu selector word to be set. Each bit in the selector covers a
490: * 'C long' in the PENDING bitfield array.
491: */
492: unsigned long evtchn_pending[sizeof(unsigned long) * 8];
493: unsigned long evtchn_mask[sizeof(unsigned long) * 8];
494:
495: /*
496: * Wallclock time: updated only by control software. Guests should base
497: * their gettimeofday() syscall on this wallclock-base value.
498: */
499: uint32_t wc_version; /* Version counter: see vcpu_time_info_t. */
500: uint32_t wc_sec; /* Secs 00:00:00 UTC, Jan 1, 1970. */
501: uint32_t wc_nsec; /* Nsecs 00:00:00 UTC, Jan 1, 1970. */
502:
503: struct arch_shared_info arch;
504:
505: };
506: #ifndef __XEN__
507: typedef struct shared_info shared_info_t;
508: #endif
509:
510: /*
511: * Start-of-day memory layout:
512: * 1. The domain is started within contiguous virtual-memory region.
513: * 2. The contiguous region ends on an aligned 4MB boundary.
514: * 3. This the order of bootstrap elements in the initial virtual region:
515: * a. relocated kernel image
516: * b. initial ram disk [mod_start, mod_len]
517: * c. list of allocated page frames [mfn_list, nr_pages]
518: * d. start_info_t structure [register ESI (x86)]
519: * e. bootstrap page tables [pt_base, CR3 (x86)]
520: * f. bootstrap stack [register ESP (x86)]
521: * 4. Bootstrap elements are packed together, but each is 4kB-aligned.
522: * 5. The initial ram disk may be omitted.
523: * 6. The list of page frames forms a contiguous 'pseudo-physical' memory
524: * layout for the domain. In particular, the bootstrap virtual-memory
525: * region is a 1:1 mapping to the first section of the pseudo-physical map.
526: * 7. All bootstrap elements are mapped read-writable for the guest OS. The
527: * only exception is the bootstrap page table, which is mapped read-only.
528: * 8. There is guaranteed to be at least 512kB padding after the final
529: * bootstrap element. If necessary, the bootstrap virtual region is
530: * extended by an extra 4MB to ensure this.
531: */
532:
533: #define MAX_GUEST_CMDLINE 1024
534: struct start_info {
535: /* THE FOLLOWING ARE FILLED IN BOTH ON INITIAL BOOT AND ON RESUME. */
536: char magic[32]; /* "xen-<version>-<platform>". */
537: unsigned long nr_pages; /* Total pages allocated to this domain. */
538: unsigned long shared_info; /* MACHINE address of shared info struct. */
539: uint32_t flags; /* SIF_xxx flags. */
540: xen_pfn_t store_mfn; /* MACHINE page number of shared page. */
541: uint32_t store_evtchn; /* Event channel for store communication. */
542: union {
543: struct {
544: xen_pfn_t mfn; /* MACHINE page number of console page. */
545: uint32_t evtchn; /* Event channel for console page. */
546: } domU;
547: struct {
548: uint32_t info_off; /* Offset of console_info struct. */
549: uint32_t info_size; /* Size of console_info struct from start.*/
550: } dom0;
551: } console;
552: /* THE FOLLOWING ARE ONLY FILLED IN ON INITIAL BOOT (NOT RESUME). */
553: unsigned long pt_base; /* VIRTUAL address of page directory. */
554: unsigned long nr_pt_frames; /* Number of bootstrap p.t. frames. */
555: unsigned long mfn_list; /* VIRTUAL address of page-frame list. */
556: unsigned long mod_start; /* VIRTUAL address of pre-loaded module. */
557: unsigned long mod_len; /* Size (bytes) of pre-loaded module. */
558: int8_t cmd_line[MAX_GUEST_CMDLINE];
559:
560: /* hackish, for multiboot compatibility */
561: unsigned mods_count;
562: };
563: typedef struct start_info start_info_t;
564:
565: /* New console union for dom0 introduced in 0x00030203. */
566: #if __XEN_INTERFACE_VERSION__ < 0x00030203
567: #define console_mfn console.domU.mfn
568: #define console_evtchn console.domU.evtchn
569: #endif
570:
571: /* These flags are passed in the 'flags' field of start_info_t. */
572: #define SIF_PRIVILEGED (1<<0) /* Is the domain privileged? */
573: #define SIF_INITDOMAIN (1<<1) /* Is this the initial control domain? */
574: #define SIF_MULTIBOOT_MOD (1<<2) /* Is this the initial control domain? */
575: #define SIF_PM_MASK (0xFF<<8) /* reserve 1 byte for xen-pm options */
576:
577: typedef struct dom0_vga_console_info {
578: uint8_t video_type; /* DOM0_VGA_CONSOLE_??? */
579: #define XEN_VGATYPE_TEXT_MODE_3 0x03
580: #define XEN_VGATYPE_VESA_LFB 0x23
581:
582: union {
583: struct {
584: /* Font height, in pixels. */
585: uint16_t font_height;
586: /* Cursor location (column, row). */
587: uint16_t cursor_x, cursor_y;
588: /* Number of rows and columns (dimensions in characters). */
589: uint16_t rows, columns;
590: } text_mode_3;
591:
592: struct {
593: /* Width and height, in pixels. */
594: uint16_t width, height;
595: /* Bytes per scan line. */
596: uint16_t bytes_per_line;
597: /* Bits per pixel. */
598: uint16_t bits_per_pixel;
599: /* LFB physical address, and size (in units of 64kB). */
600: uint32_t lfb_base;
601: uint32_t lfb_size;
602: /* RGB mask offsets and sizes, as defined by VBE 1.2+ */
603: uint8_t red_pos, red_size;
604: uint8_t green_pos, green_size;
605: uint8_t blue_pos, blue_size;
606: uint8_t rsvd_pos, rsvd_size;
607: #if __XEN_INTERFACE_VERSION__ >= 0x00030206
608: /* VESA capabilities (offset 0xa, VESA command 0x4f00). */
609: uint32_t gbl_caps;
610: /* Mode attributes (offset 0x0, VESA command 0x4f01). */
611: uint16_t mode_attrs;
612: #endif
613: } vesa_lfb;
614: } u;
615: } dom0_vga_console_info_t;
616: #define xen_vga_console_info dom0_vga_console_info
617: #define xen_vga_console_info_t dom0_vga_console_info_t
618:
619: typedef uint8_t xen_domain_handle_t[16];
620:
621: /* Turn a plain number into a C unsigned long constant. */
622: #define __mk_unsigned_long(x) x ## UL
623: #define mk_unsigned_long(x) __mk_unsigned_long(x)
624:
625: __DEFINE_XEN_GUEST_HANDLE(uint8, uint8_t);
626: __DEFINE_XEN_GUEST_HANDLE(uint16, uint16_t);
627: __DEFINE_XEN_GUEST_HANDLE(uint32, uint32_t);
628: __DEFINE_XEN_GUEST_HANDLE(uint64, uint64_t);
629:
630: #else /* __ASSEMBLY__ */
631:
632: /* In assembly code we cannot use C numeric constant suffixes. */
633: #define mk_unsigned_long(x) x
634:
635: #endif /* !__ASSEMBLY__ */
636:
637: /* Default definitions for macros used by domctl/sysctl. */
638: #if defined(__XEN__) || defined(__XEN_TOOLS__)
639: #ifndef uint64_aligned_t
640: #define uint64_aligned_t uint64_t
641: #endif
642: #ifndef XEN_GUEST_HANDLE_64
643: #define XEN_GUEST_HANDLE_64(name) XEN_GUEST_HANDLE(name)
644: #endif
645: #endif
646:
647: #endif /* __XEN_PUBLIC_XEN_H__ */
648:
649: /*
650: * Local variables:
651: * mode: C
652: * c-set-style: "BSD"
653: * c-basic-offset: 4
654: * tab-width: 4
655: * indent-tabs-mode: nil
656: * End:
657: */
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.