sys/arch/xen/include/xen3-public/xen.h

   1 /* $NetBSD: xen.h,v 1.8 2008/05/04 19:56:28 cegger Exp $ */
   2 /******************************************************************************
   3  * xen.h
   4  *
   5  * Guest OS interface to Xen.
   6  *
   7  * Permission is hereby granted, free of charge, to any person obtaining a copy
   8  * of this software and associated documentation files (the "Software"), to
   9  * deal in the Software without restriction, including without limitation the
  10  * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
  11  * sell copies of the Software, and to permit persons to whom the Software is
  12  * furnished to do so, subject to the following conditions:
  13  *
  14  * The above copyright notice and this permission notice shall be included in
  15  * all copies or substantial portions of the Software.
  16  *
  17  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  18  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  19  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  20  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  21  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  22  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
  23  * DEALINGS IN THE SOFTWARE.
  24  *
  25  * Copyright (c) 2004, K A Fraser
  26  */
  27
  28 #ifndef __XEN_PUBLIC_XEN_H__
  29 #define __XEN_PUBLIC_XEN_H__
  30
  31 #include "xen-compat.h"
  32
  33 #if defined(__i386__) || defined(__x86_64__)
  34 #include "arch-x86/xen.h"
  35 #elif defined(__ia64__)
  36 #include "arch-ia64.h"
  37 #else
  38 #error "Unsupported architecture"
  39 #endif
  40
  41 #ifndef __ASSEMBLY__
  42 /* Guest handles for primitive C types. */
  43 DEFINE_XEN_GUEST_HANDLE(char);
  44 __DEFINE_XEN_GUEST_HANDLE(uchar, unsigned char);
  45 DEFINE_XEN_GUEST_HANDLE(int);
  46 __DEFINE_XEN_GUEST_HANDLE(uint,  unsigned int);
  47 DEFINE_XEN_GUEST_HANDLE(long);
  48 __DEFINE_XEN_GUEST_HANDLE(ulong, unsigned long);
  49 DEFINE_XEN_GUEST_HANDLE(void);
  50
  51 DEFINE_XEN_GUEST_HANDLE(xen_pfn_t);
  52 #endif
  53
  54 /*
  55  * HYPERCALLS
  56  */
  57
  58 #define __HYPERVISOR_set_trap_table        0
  59 #define __HYPERVISOR_mmu_update            1
  60 #define __HYPERVISOR_set_gdt               2
  61 #define __HYPERVISOR_stack_switch          3
  62 #define __HYPERVISOR_set_callbacks         4
  63 #define __HYPERVISOR_fpu_taskswitch        5
  64 #define __HYPERVISOR_sched_op_compat       6 /* compat since 0x00030101 */
  65 #define __HYPERVISOR_platform_op           7
  66 #define __HYPERVISOR_set_debugreg          8
  67 #define __HYPERVISOR_get_debugreg          9
  68 #define __HYPERVISOR_update_descriptor    10
  69 #define __HYPERVISOR_memory_op            12
  70 #define __HYPERVISOR_multicall            13
  71 #define __HYPERVISOR_update_va_mapping    14
  72 #define __HYPERVISOR_set_timer_op         15
  73 #define __HYPERVISOR_event_channel_op_compat 16 /* compat since 0x00030202 */
  74 #define __HYPERVISOR_xen_version          17
  75 #define __HYPERVISOR_console_io           18
  76 #define __HYPERVISOR_physdev_op_compat    19 /* compat since 0x00030202 */
  77 #define __HYPERVISOR_grant_table_op       20
  78 #define __HYPERVISOR_vm_assist            21
  79 #define __HYPERVISOR_update_va_mapping_otherdomain 22
  80 #define __HYPERVISOR_iret                 23 /* x86 only */
  81 #define __HYPERVISOR_vcpu_op              24
  82 #define __HYPERVISOR_set_segment_base     25 /* x86/64 only */
  83 #define __HYPERVISOR_mmuext_op            26
  84 #define __HYPERVISOR_xsm_op               27
  85 #define __HYPERVISOR_nmi_op               28
  86 #define __HYPERVISOR_sched_op             29
  87 #define __HYPERVISOR_callback_op          30
  88 #define __HYPERVISOR_xenoprof_op          31
  89 #define __HYPERVISOR_event_channel_op     32
  90 #define __HYPERVISOR_physdev_op           33
  91 #define __HYPERVISOR_hvm_op               34
  92 #define __HYPERVISOR_sysctl               35
  93 #define __HYPERVISOR_domctl               36
  94 #define __HYPERVISOR_kexec_op             37
  95
  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
 105
 106 /*
 107  * HYPERCALL COMPATIBILITY.
 108  */
 109
 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
 115
 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
 123
 124 /* New platform_op hypercall introduced in 0x00030204. */
 125 #if __XEN_INTERFACE_VERSION__ < 0x00030204
 126 #define __HYPERVISOR_dom0_op __HYPERVISOR_platform_op
 127 #endif
 128
 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            */
 147
 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
 157
 158 #define NR_VIRQS       24
 159
 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)) */
 191
 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 #define MMUEXT_PIN_L1_TABLE      0
 237 #define MMUEXT_PIN_L2_TABLE      1
 238 #define MMUEXT_PIN_L3_TABLE      2
 239 #define MMUEXT_PIN_L4_TABLE      3
 240 #define MMUEXT_UNPIN_TABLE       4
 241 #define MMUEXT_NEW_BASEPTR       5
 242 #define MMUEXT_TLB_FLUSH_LOCAL   6
 243 #define MMUEXT_INVLPG_LOCAL      7
 244 #define MMUEXT_TLB_FLUSH_MULTI   8
 245 #define MMUEXT_INVLPG_MULTI      9
 246 #define MMUEXT_TLB_FLUSH_ALL    10
 247 #define MMUEXT_INVLPG_ALL       11
 248 #define MMUEXT_FLUSH_CACHE      12
 249 #define MMUEXT_SET_LDT          13
 250 #define MMUEXT_NEW_USER_BASEPTR 15
 251
 252 #ifndef __ASSEMBLY__
 253 struct mmuext_op {
 254     unsigned int cmd;
 255     union {
 256         /* [UN]PIN_TABLE, NEW_BASEPTR, NEW_USER_BASEPTR */
 257         xen_pfn_t     mfn;
 258         /* INVLPG_LOCAL, INVLPG_ALL, SET_LDT */
 259         unsigned long linear_addr;
 260     } arg1;
 261     union {
 262         /* SET_LDT */
 263         unsigned int nr_ents;
 264         /* TLB_FLUSH_MULTI, INVLPG_MULTI */
 265 #if __XEN_INTERFACE_VERSION__ >= 0x00030205
 266         XEN_GUEST_HANDLE(void) vcpumask;
 267 #else
 268         void *vcpumask;
 269 #endif
 270     } arg2;
 271 };
 272 typedef struct mmuext_op mmuext_op_t;
 273 DEFINE_XEN_GUEST_HANDLE(mmuext_op_t);
 274 #endif
 275
 276 /* These are passed as 'flags' to update_va_mapping. They can be ORed. */
 277 /* When specifying UVMF_MULTI, also OR in a pointer to a CPU bitmap.   */
 278 /* UVMF_LOCAL is merely UVMF_MULTI with a NULL bitmap pointer.         */
 279 #define UVMF_NONE               (0UL<<0) /* No flushing at all.   */
 280 #define UVMF_TLB_FLUSH          (1UL<<0) /* Flush entire TLB(s).  */
 281 #define UVMF_INVLPG             (2UL<<0) /* Flush only one entry. */
 282 #define UVMF_FLUSHTYPE_MASK     (3UL<<0)
 283 #define UVMF_MULTI              (0UL<<2) /* Flush subset of TLBs. */
 284 #define UVMF_LOCAL              (0UL<<2) /* Flush local TLB.      */
 285 #define UVMF_ALL                (1UL<<2) /* Flush all TLBs.       */
 286
 287 /*
 288  * Commands to HYPERVISOR_console_io().
 289  */
 290 #define CONSOLEIO_write         0
 291 #define CONSOLEIO_read          1
 292
 293 /*
 294  * Commands to HYPERVISOR_vm_assist().
 295  */
 296 #define VMASST_CMD_enable                0
 297 #define VMASST_CMD_disable               1
 298
 299 /* x86/32 guests: simulate full 4GB segment limits. */
 300 #define VMASST_TYPE_4gb_segments         0
 301
 302 /* x86/32 guests: trap (vector 15) whenever above vmassist is used. */
 303 #define VMASST_TYPE_4gb_segments_notify  1
 304
 305 /*
 306  * x86 guests: support writes to bottom-level PTEs.
 307  * NB1. Page-directory entries cannot be written.
 308  * NB2. Guest must continue to remove all writable mappings of PTEs.
 309  */
 310 #define VMASST_TYPE_writable_pagetables  2
 311
 312 /* x86/PAE guests: support PDPTs above 4GB. */
 313 #define VMASST_TYPE_pae_extended_cr3     3
 314
 315 #define MAX_VMASST_TYPE                  3
 316
 317 #ifndef __ASSEMBLY__
 318
 319 typedef uint16_t domid_t;
 320
 321 /* Domain ids >= DOMID_FIRST_RESERVED cannot be used for ordinary domains. */
 322 #define DOMID_FIRST_RESERVED (0x7FF0U)
 323
 324 /* DOMID_SELF is used in certain contexts to refer to oneself. */
 325 #define DOMID_SELF (0x7FF0U)
 326
 327 /*
 328  * DOMID_IO is used to restrict page-table updates to mapping I/O memory.
 329  * Although no Foreign Domain need be specified to map I/O pages, DOMID_IO
 330  * is useful to ensure that no mappings to the OS's own heap are accidentally
 331  * installed. (e.g., in Linux this could cause havoc as reference counts
 332  * aren't adjusted on the I/O-mapping code path).
 333  * This only makes sense in MMUEXT_SET_FOREIGNDOM, but in that context can
 334  * be specified by any calling domain.
 335  */
 336 #define DOMID_IO   (0x7FF1U)
 337
 338 /*
 339  * DOMID_XEN is used to allow privileged domains to map restricted parts of
 340  * Xen's heap space (e.g., the machine_to_phys table).
 341  * This only makes sense in MMUEXT_SET_FOREIGNDOM, and is only permitted if
 342  * the caller is privileged.
 343  */
 344 #define DOMID_XEN  (0x7FF2U)
 345
 346 /*
 347  * Send an array of these to HYPERVISOR_mmu_update().
 348  * NB. The fields are natural pointer/address size for this architecture.
 349  */
 350 struct mmu_update {
 351     uint64_t ptr;       /* Machine address of PTE. */
 352     uint64_t val;       /* New contents of PTE.    */
 353 };
 354 typedef struct mmu_update mmu_update_t;
 355 DEFINE_XEN_GUEST_HANDLE(mmu_update_t);
 356
 357 /*
 358  * Send an array of these to HYPERVISOR_multicall().
 359  * NB. The fields are natural register size for this architecture.
 360  */
 361 struct multicall_entry {
 362     unsigned long op, result;
 363     unsigned long args[6];
 364 };
 365 typedef struct multicall_entry multicall_entry_t;
 366 DEFINE_XEN_GUEST_HANDLE(multicall_entry_t);
 367
 368 /*
 369  * Event channel endpoints per domain:
 370  *  1024 if a long is 32 bits; 4096 if a long is 64 bits.
 371  */
 372 #define NR_EVENT_CHANNELS (sizeof(unsigned long) * sizeof(unsigned long) * 64)
 373
 374 struct vcpu_time_info {
 375     /*
 376      * Updates to the following values are preceded and followed by an
 377      * increment of 'version'. The guest can therefore detect updates by
 378      * looking for changes to 'version'. If the least-significant bit of
 379      * the version number is set then an update is in progress and the guest
 380      * must wait to read a consistent set of values.
 381      * The correct way to interact with the version number is similar to
 382      * Linux's seqlock: see the implementations of read_seqbegin/read_seqretry.
 383      */
 384     uint32_t version;
 385     uint32_t pad0;
 386     uint64_t tsc_timestamp;   /* TSC at last update of time vals.  */
 387     uint64_t system_time;     /* Time, in nanosecs, since boot.    */
 388     /*
 389      * Current system time:
 390      *   system_time +
 391      *   ((((tsc - tsc_timestamp) << tsc_shift) * tsc_to_system_mul) >> 32)
 392      * CPU frequency (Hz):
 393      *   ((10^9 << 32) / tsc_to_system_mul) >> tsc_shift
 394      */
 395     uint32_t tsc_to_system_mul;
 396     int8_t   tsc_shift;
 397     int8_t   pad1[3];
 398 }; /* 32 bytes */
 399 typedef struct vcpu_time_info vcpu_time_info_t;
 400
 401 struct vcpu_info {
 402     /*
 403      * 'evtchn_upcall_pending' is written non-zero by Xen to indicate
 404      * a pending notification for a particular VCPU. It is then cleared
 405      * by the guest OS /before/ checking for pending work, thus avoiding
 406      * a set-and-check race. Note that the mask is only accessed by Xen
 407      * on the CPU that is currently hosting the VCPU. This means that the
 408      * pending and mask flags can be updated by the guest without special
 409      * synchronisation (i.e., no need for the x86 LOCK prefix).
 410      * This may seem suboptimal because if the pending flag is set by
 411      * a different CPU then an IPI may be scheduled even when the mask
 412      * is set. However, note:
 413      *  1. The task of 'interrupt holdoff' is covered by the per-event-
 414      *     channel mask bits. A 'noisy' event that is continually being
 415      *     triggered can be masked at source at this very precise
 416      *     granularity.
 417      *  2. The main purpose of the per-VCPU mask is therefore to restrict
 418      *     reentrant execution: whether for concurrency control, or to
 419      *     prevent unbounded stack usage. Whatever the purpose, we expect
 420      *     that the mask will be asserted only for short periods at a time,
 421      *     and so the likelihood of a 'spurious' IPI is suitably small.
 422      * The mask is read before making an event upcall to the guest: a
 423      * non-zero mask therefore guarantees that the VCPU will not receive
 424      * an upcall activation. The mask is cleared when the VCPU requests
 425      * to block: this avoids wakeup-waiting races.
 426      */
 427     uint8_t evtchn_upcall_pending;
 428     uint8_t evtchn_upcall_mask;
 429     unsigned long evtchn_pending_sel;
 430     struct arch_vcpu_info arch;
 431     struct vcpu_time_info time;
 432 }; /* 64 bytes (x86) */
 433 #ifndef __XEN__
 434 typedef struct vcpu_info vcpu_info_t;
 435 #endif
 436
 437 /*
 438  * Xen/kernel shared data -- pointer provided in start_info.
 439  *
 440  * This structure is defined to be both smaller than a page, and the
 441  * only data on the shared page, but may vary in actual size even within
 442  * compatible Xen versions; guests should not rely on the size
 443  * of this structure remaining constant.
 444  */
 445 struct shared_info {
 446     struct vcpu_info vcpu_info[MAX_VIRT_CPUS];
 447
 448     /*
 449      * A domain can create "event channels" on which it can send and receive
 450      * asynchronous event notifications. There are three classes of event that
 451      * are delivered by this mechanism:
 452      *  1. Bi-directional inter- and intra-domain connections. Domains must
 453      *     arrange out-of-band to set up a connection (usually by allocating
 454      *     an unbound 'listener' port and avertising that via a storage service
 455      *     such as xenstore).
 456      *  2. Physical interrupts. A domain with suitable hardware-access
 457      *     privileges can bind an event-channel port to a physical interrupt
 458      *     source.
 459      *  3. Virtual interrupts ('events'). A domain can bind an event-channel
 460      *     port to a virtual interrupt source, such as the virtual-timer
 461      *     device or the emergency console.
 462      *
 463      * Event channels are addressed by a "port index". Each channel is
 464      * associated with two bits of information:
 465      *  1. PENDING -- notifies the domain that there is a pending notification
 466      *     to be processed. This bit is cleared by the guest.
 467      *  2. MASK -- if this bit is clear then a 0->1 transition of PENDING
 468      *     will cause an asynchronous upcall to be scheduled. This bit is only
 469      *     updated by the guest. It is read-only within Xen. If a channel
 470      *     becomes pending while the channel is masked then the 'edge' is lost
 471      *     (i.e., when the channel is unmasked, the guest must manually handle
 472      *     pending notifications as no upcall will be scheduled by Xen).
 473      *
 474      * To expedite scanning of pending notifications, any 0->1 pending
 475      * transition on an unmasked channel causes a corresponding bit in a
 476      * per-vcpu selector word to be set. Each bit in the selector covers a
 477      * 'C long' in the PENDING bitfield array.
 478      */
 479     unsigned long evtchn_pending[sizeof(unsigned long) * 8];
 480     unsigned long evtchn_mask[sizeof(unsigned long) * 8];
 481
 482     /*
 483      * Wallclock time: updated only by control software. Guests should base
 484      * their gettimeofday() syscall on this wallclock-base value.
 485      */
 486     uint32_t wc_version;      /* Version counter: see vcpu_time_info_t. */
 487     uint32_t wc_sec;          /* Secs  00:00:00 UTC, Jan 1, 1970.  */
 488     uint32_t wc_nsec;         /* Nsecs 00:00:00 UTC, Jan 1, 1970.  */
 489
 490     struct arch_shared_info arch;
 491
 492 };
 493 #ifndef __XEN__
 494 typedef struct shared_info shared_info_t;
 495 #endif
 496
 497 /*
 498  * Start-of-day memory layout:
 499  *  1. The domain is started within contiguous virtual-memory region.
 500  *  2. The contiguous region ends on an aligned 4MB boundary.
 501  *  3. This the order of bootstrap elements in the initial virtual region:
 502  *      a. relocated kernel image
 503  *      b. initial ram disk              [mod_start, mod_len]
 504  *      c. list of allocated page frames [mfn_list, nr_pages]
 505  *      d. start_info_t structure        [register ESI (x86)]
 506  *      e. bootstrap page tables         [pt_base, CR3 (x86)]
 507  *      f. bootstrap stack               [register ESP (x86)]
 508  *  4. Bootstrap elements are packed together, but each is 4kB-aligned.
 509  *  5. The initial ram disk may be omitted.
 510  *  6. The list of page frames forms a contiguous 'pseudo-physical' memory
 511  *     layout for the domain. In particular, the bootstrap virtual-memory
 512  *     region is a 1:1 mapping to the first section of the pseudo-physical map.
 513  *  7. All bootstrap elements are mapped read-writable for the guest OS. The
 514  *     only exception is the bootstrap page table, which is mapped read-only.
 515  *  8. There is guaranteed to be at least 512kB padding after the final
 516  *     bootstrap element. If necessary, the bootstrap virtual region is
 517  *     extended by an extra 4MB to ensure this.
 518  */
 519
 520 #define MAX_GUEST_CMDLINE 1024
 521 struct start_info {
 522     /* THE FOLLOWING ARE FILLED IN BOTH ON INITIAL BOOT AND ON RESUME.    */
 523     char magic[32];             /* "xen-<version>-<platform>".            */
 524     unsigned long nr_pages;     /* Total pages allocated to this domain.  */
 525     unsigned long shared_info;  /* MACHINE address of shared info struct. */
 526     uint32_t flags;             /* SIF_xxx flags.                         */
 527     xen_pfn_t store_mfn;        /* MACHINE page number of shared page.    */
 528     uint32_t store_evtchn;      /* Event channel for store communication. */
 529     union {
 530         struct {
 531             xen_pfn_t mfn;      /* MACHINE page number of console page.   */
 532             uint32_t  evtchn;   /* Event channel for console page.        */
 533         } domU;
 534         struct {
 535             uint32_t info_off;  /* Offset of console_info struct.         */
 536             uint32_t info_size; /* Size of console_info struct from start.*/
 537         } dom0;
 538     } console;
 539     /* THE FOLLOWING ARE ONLY FILLED IN ON INITIAL BOOT (NOT RESUME).     */
 540     unsigned long pt_base;      /* VIRTUAL address of page directory.     */
 541     unsigned long nr_pt_frames; /* Number of bootstrap p.t. frames.       */
 542     unsigned long mfn_list;     /* VIRTUAL address of page-frame list.    */
 543     unsigned long mod_start;    /* VIRTUAL address of pre-loaded module.  */
 544     unsigned long mod_len;      /* Size (bytes) of pre-loaded module.     */
 545     int8_t cmd_line[MAX_GUEST_CMDLINE];
 546 };
 547 typedef struct start_info start_info_t;
 548
 549 /* New console union for dom0 introduced in 0x00030203. */
 550 #if __XEN_INTERFACE_VERSION__ < 0x00030203
 551 #define console_mfn    console.domU.mfn
 552 #define console_evtchn console.domU.evtchn
 553 #endif
 554
 555 /* These flags are passed in the 'flags' field of start_info_t. */
 556 #define SIF_PRIVILEGED    (1<<0)  /* Is the domain privileged? */
 557 #define SIF_INITDOMAIN    (1<<1)  /* Is this the initial control domain? */
 558 #define SIF_PM_MASK       (0xFF<<8) /* reserve 1 byte for xen-pm options */
 559
 560 typedef struct dom0_vga_console_info {
 561     uint8_t video_type; /* DOM0_VGA_CONSOLE_??? */
 562 #define XEN_VGATYPE_TEXT_MODE_3 0x03
 563 #define XEN_VGATYPE_VESA_LFB    0x23
 564
 565     union {
 566         struct {
 567             /* Font height, in pixels. */
 568             uint16_t font_height;
 569             /* Cursor location (column, row). */
 570             uint16_t cursor_x, cursor_y;
 571             /* Number of rows and columns (dimensions in characters). */
 572             uint16_t rows, columns;
 573         } text_mode_3;
 574
 575         struct {
 576             /* Width and height, in pixels. */
 577             uint16_t width, height;
 578             /* Bytes per scan line. */
 579             uint16_t bytes_per_line;
 580             /* Bits per pixel. */
 581             uint16_t bits_per_pixel;
 582             /* LFB physical address, and size (in units of 64kB). */
 583             uint32_t lfb_base;
 584             uint32_t lfb_size;
 585             /* RGB mask offsets and sizes, as defined by VBE 1.2+ */
 586             uint8_t  red_pos, red_size;
 587             uint8_t  green_pos, green_size;
 588             uint8_t  blue_pos, blue_size;
 589             uint8_t  rsvd_pos, rsvd_size;
 590 #if __XEN_INTERFACE_VERSION__ >= 0x00030206
 591             /* VESA capabilities (offset 0xa, VESA command 0x4f00). */
 592             uint32_t gbl_caps;
 593             /* Mode attributes (offset 0x0, VESA command 0x4f01). */
 594             uint16_t mode_attrs;
 595 #endif
 596         } vesa_lfb;
 597     } u;
 598 } dom0_vga_console_info_t;
 599 #define xen_vga_console_info dom0_vga_console_info
 600 #define xen_vga_console_info_t dom0_vga_console_info_t
 601
 602 typedef uint8_t xen_domain_handle_t[16];
 603
 604 /* Turn a plain number into a C unsigned long constant. */
 605 #define __mk_unsigned_long(x) x ## UL
 606 #define mk_unsigned_long(x) __mk_unsigned_long(x)
 607
 608 __DEFINE_XEN_GUEST_HANDLE(uint8,  uint8_t);
 609 __DEFINE_XEN_GUEST_HANDLE(uint16, uint16_t);
 610 __DEFINE_XEN_GUEST_HANDLE(uint32, uint32_t);
 611 __DEFINE_XEN_GUEST_HANDLE(uint64, uint64_t);
 612
 613 #else /* __ASSEMBLY__ */
 614
 615 /* In assembly code we cannot use C numeric constant suffixes. */
 616 #define mk_unsigned_long(x) x
 617
 618 #endif /* !__ASSEMBLY__ */
 619
 620 /* Default definitions for macros used by domctl/sysctl. */
 621 #if defined(__XEN__) || defined(__XEN_TOOLS__)
 622 #ifndef uint64_aligned_t
 623 #define uint64_aligned_t uint64_t
 624 #endif
 625 #ifndef XEN_GUEST_HANDLE_64
 626 #define XEN_GUEST_HANDLE_64(name) XEN_GUEST_HANDLE(name)
 627 #endif
 628 #endif
 629
 630 #endif /* __XEN_PUBLIC_XEN_H__ */
 631
 632 /*
 633  * Local variables:
 634  * mode: C
 635  * c-set-style: "BSD"
 636  * c-basic-offset: 4
 637  * tab-width: 4
 638  * indent-tabs-mode: nil
 639  * End:
 640  */