1/******************************************************************************
2 * arch-x86/xen.h
3 *
4 * Guest OS interface to x86 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) 2004-2006, K A Fraser
25 */
26
27#include "../xen.h"
28
29#ifndef __XEN_PUBLIC_ARCH_X86_XEN_H__
30#define __XEN_PUBLIC_ARCH_X86_XEN_H__
31
32/* Structural guest handles introduced in 0x00030201. */
33#if __XEN_INTERFACE_VERSION__ >= 0x00030201
34#define ___DEFINE_XEN_GUEST_HANDLE(name, type) \
35    typedef struct { type *p; } __guest_handle_ ## name
36#else
37#define ___DEFINE_XEN_GUEST_HANDLE(name, type) \
38    typedef type * __guest_handle_ ## name
39#endif
40
41/*
42 * XEN_GUEST_HANDLE represents a guest pointer, when passed as a field
43 * in a struct in memory.
44 * XEN_GUEST_HANDLE_PARAM represent a guest pointer, when passed as an
45 * hypercall argument.
46 * XEN_GUEST_HANDLE_PARAM and XEN_GUEST_HANDLE are the same on X86 but
47 * they might not be on other architectures.
48 */
49#define __DEFINE_XEN_GUEST_HANDLE(name, type) \
50    ___DEFINE_XEN_GUEST_HANDLE(name, type);   \
51    ___DEFINE_XEN_GUEST_HANDLE(const_##name, const type)
52#define DEFINE_XEN_GUEST_HANDLE(name)   __DEFINE_XEN_GUEST_HANDLE(name, name)
53#define __XEN_GUEST_HANDLE(name)        __guest_handle_ ## name
54#define XEN_GUEST_HANDLE(name)          __XEN_GUEST_HANDLE(name)
55#define XEN_GUEST_HANDLE_PARAM(name)    XEN_GUEST_HANDLE(name)
56#define set_xen_guest_handle_raw(hnd, val)  do { (hnd).p = val; } while (0)
57#define set_xen_guest_handle(hnd, val) set_xen_guest_handle_raw(hnd, val)
58
59#if defined(__i386__)
60# ifdef __XEN__
61__DeFiNe__ __DECL_REG_LO8(which) uint32_t e ## which ## x
62__DeFiNe__ __DECL_REG_LO16(name) union { uint32_t e ## name; }
63# endif
64#include "xen-x86_32.h"
65# ifdef __XEN__
66__UnDeF__ __DECL_REG_LO8
67__UnDeF__ __DECL_REG_LO16
68__DeFiNe__ __DECL_REG_LO8(which) e ## which ## x
69__DeFiNe__ __DECL_REG_LO16(name) e ## name
70# endif
71#elif defined(__x86_64__)
72#include "xen-x86_64.h"
73#endif
74
75#ifndef __ASSEMBLY__
76typedef unsigned long xen_pfn_t;
77#define PRI_xen_pfn "lx"
78#define PRIu_xen_pfn "lu"
79#endif
80
81#define XEN_HAVE_PV_GUEST_ENTRY 1
82
83#define XEN_HAVE_PV_UPCALL_MASK 1
84
85/*
86 * `incontents 200 segdesc Segment Descriptor Tables
87 */
88/*
89 * ` enum neg_errnoval
90 * ` HYPERVISOR_set_gdt(const xen_pfn_t frames[], unsigned int entries);
91 * `
92 */
93/*
94 * A number of GDT entries are reserved by Xen. These are not situated at the
95 * start of the GDT because some stupid OSes export hard-coded selector values
96 * in their ABI. These hard-coded values are always near the start of the GDT,
97 * so Xen places itself out of the way, at the far end of the GDT.
98 *
99 * NB The LDT is set using the MMUEXT_SET_LDT op of HYPERVISOR_mmuext_op
100 */
101#define FIRST_RESERVED_GDT_PAGE  14
102#define FIRST_RESERVED_GDT_BYTE  (FIRST_RESERVED_GDT_PAGE * 4096)
103#define FIRST_RESERVED_GDT_ENTRY (FIRST_RESERVED_GDT_BYTE / 8)
104
105
106/*
107 * ` enum neg_errnoval
108 * ` HYPERVISOR_update_descriptor(u64 pa, u64 desc);
109 * `
110 * ` @pa   The machine physical address of the descriptor to
111 * `       update. Must be either a descriptor page or writable.
112 * ` @desc The descriptor value to update, in the same format as a
113 * `       native descriptor table entry.
114 */
115
116/* Maximum number of virtual CPUs in legacy multi-processor guests. */
117#define XEN_LEGACY_MAX_VCPUS 32
118
119#ifndef __ASSEMBLY__
120
121typedef unsigned long xen_ulong_t;
122#define PRI_xen_ulong "lx"
123
124/*
125 * ` enum neg_errnoval
126 * ` HYPERVISOR_stack_switch(unsigned long ss, unsigned long esp);
127 * `
128 * Sets the stack segment and pointer for the current vcpu.
129 */
130
131/*
132 * ` enum neg_errnoval
133 * ` HYPERVISOR_set_trap_table(const struct trap_info traps[]);
134 * `
135 */
136/*
137 * Send an array of these to HYPERVISOR_set_trap_table().
138 * Terminate the array with a sentinel entry, with traps[].address==0.
139 * The privilege level specifies which modes may enter a trap via a software
140 * interrupt. On x86/64, since rings 1 and 2 are unavailable, we allocate
141 * privilege levels as follows:
142 *  Level == 0: Noone may enter
143 *  Level == 1: Kernel may enter
144 *  Level == 2: Kernel may enter
145 *  Level == 3: Everyone may enter
146 */
147#define TI_GET_DPL(_ti)      ((_ti)->flags & 3)
148#define TI_GET_IF(_ti)       ((_ti)->flags & 4)
149#define TI_SET_DPL(_ti,_dpl) ((_ti)->flags |= (_dpl))
150#define TI_SET_IF(_ti,_if)   ((_ti)->flags |= ((!!(_if))<<2))
151struct trap_info {
152    uint8_t       vector;  /* exception vector                              */
153    uint8_t       flags;   /* 0-3: privilege level; 4: clear event enable?  */
154    uint16_t      cs;      /* code selector                                 */
155    unsigned long address; /* code offset                                   */
156};
157typedef struct trap_info trap_info_t;
158DEFINE_XEN_GUEST_HANDLE(trap_info_t);
159
160typedef uint64_t tsc_timestamp_t; /* RDTSC timestamp */
161
162/*
163 * The following is all CPU context. Note that the fpu_ctxt block is filled
164 * in by FXSAVE if the CPU has feature FXSR; otherwise FSAVE is used.
165 *
166 * Also note that when calling DOMCTL_setvcpucontext for HVM guests, not all
167 * information in this structure is updated, the fields read include: fpu_ctxt
168 * (if VGCT_I387_VALID is set), flags, user_regs and debugreg[*].
169 *
170 * Note: VCPUOP_initialise for HVM guests is non-symetric with
171 * DOMCTL_setvcpucontext, and uses struct vcpu_hvm_context from hvm/hvm_vcpu.h
172 */
173struct vcpu_guest_context {
174    /* FPU registers come first so they can be aligned for FXSAVE/FXRSTOR. */
175    struct { char x[512]; } fpu_ctxt;       /* User-level FPU registers     */
176#define VGCF_I387_VALID                (1<<0)
177#define VGCF_IN_KERNEL                 (1<<2)
178#define _VGCF_i387_valid               0
179#define VGCF_i387_valid                (1<<_VGCF_i387_valid)
180#define _VGCF_in_kernel                2
181#define VGCF_in_kernel                 (1<<_VGCF_in_kernel)
182#define _VGCF_failsafe_disables_events 3
183#define VGCF_failsafe_disables_events  (1<<_VGCF_failsafe_disables_events)
184#define _VGCF_syscall_disables_events  4
185#define VGCF_syscall_disables_events   (1<<_VGCF_syscall_disables_events)
186#define _VGCF_online                   5
187#define VGCF_online                    (1<<_VGCF_online)
188    unsigned long flags;                    /* VGCF_* flags                 */
189    struct cpu_user_regs user_regs;         /* User-level CPU registers     */
190    struct trap_info trap_ctxt[256];        /* Virtual IDT                  */
191    unsigned long ldt_base, ldt_ents;       /* LDT (linear address, # ents) */
192    unsigned long gdt_frames[16], gdt_ents; /* GDT (machine frames, # ents) */
193    unsigned long kernel_ss, kernel_sp;     /* Virtual TSS (only SS1/SP1)   */
194    /* NB. User pagetable on x86/64 is placed in ctrlreg[1]. */
195    unsigned long ctrlreg[8];               /* CR0-CR7 (control registers)  */
196    unsigned long debugreg[8];              /* DB0-DB7 (debug registers)    */
197#ifdef __i386__
198    unsigned long event_callback_cs;        /* CS:EIP of event callback     */
199    unsigned long event_callback_eip;
200    unsigned long failsafe_callback_cs;     /* CS:EIP of failsafe callback  */
201    unsigned long failsafe_callback_eip;
202#else
203    unsigned long event_callback_eip;
204    unsigned long failsafe_callback_eip;
205#ifdef __XEN__
206    union {
207        unsigned long syscall_callback_eip;
208        struct {
209            unsigned int event_callback_cs;    /* compat CS of event cb     */
210            unsigned int failsafe_callback_cs; /* compat CS of failsafe cb  */
211        };
212    };
213#else
214    unsigned long syscall_callback_eip;
215#endif
216#endif
217    unsigned long vm_assist;                /* VMASST_TYPE_* bitmap */
218#ifdef __x86_64__
219    /* Segment base addresses. */
220    uint64_t      fs_base;
221    uint64_t      gs_base_kernel;
222    uint64_t      gs_base_user;
223#endif
224};
225typedef struct vcpu_guest_context vcpu_guest_context_t;
226DEFINE_XEN_GUEST_HANDLE(vcpu_guest_context_t);
227
228struct arch_shared_info {
229    /*
230     * Number of valid entries in the p2m table(s) anchored at
231     * pfn_to_mfn_frame_list_list and/or p2m_vaddr.
232     */
233    unsigned long max_pfn;
234    /*
235     * Frame containing list of mfns containing list of mfns containing p2m.
236     * A value of 0 indicates it has not yet been set up, ~0 indicates it has
237     * been set to invalid e.g. due to the p2m being too large for the 3-level
238     * p2m tree. In this case the linear mapper p2m list anchored at p2m_vaddr
239     * is to be used.
240     */
241    xen_pfn_t     pfn_to_mfn_frame_list_list;
242    unsigned long nmi_reason;
243    /*
244     * Following three fields are valid if p2m_cr3 contains a value different
245     * from 0.
246     * p2m_cr3 is the root of the address space where p2m_vaddr is valid.
247     * p2m_cr3 is in the same format as a cr3 value in the vcpu register state
248     * and holds the folded machine frame number (via xen_pfn_to_cr3) of a
249     * L3 or L4 page table.
250     * p2m_vaddr holds the virtual address of the linear p2m list. All entries
251     * in the range [0...max_pfn[ are accessible via this pointer.
252     * p2m_generation will be incremented by the guest before and after each
253     * change of the mappings of the p2m list. p2m_generation starts at 0 and
254     * a value with the least significant bit set indicates that a mapping
255     * update is in progress. This allows guest external software (e.g. in Dom0)
256     * to verify that read mappings are consistent and whether they have changed
257     * since the last check.
258     * Modifying a p2m element in the linear p2m list is allowed via an atomic
259     * write only.
260     */
261    unsigned long p2m_cr3;         /* cr3 value of the p2m address space */
262    unsigned long p2m_vaddr;       /* virtual address of the p2m list */
263    unsigned long p2m_generation;  /* generation count of p2m mapping */
264#ifdef __i386__
265    /* There's no room for this field in the generic structure. */
266    uint32_t wc_sec_hi;
267#endif
268};
269typedef struct arch_shared_info arch_shared_info_t;
270
271#if defined(__XEN__) || defined(__XEN_TOOLS__)
272/*
273 * struct xen_arch_domainconfig's ABI is covered by
274 * XEN_DOMCTL_INTERFACE_VERSION.
275 */
276struct xen_arch_domainconfig {
277#define _XEN_X86_EMU_LAPIC          0
278#define XEN_X86_EMU_LAPIC           (1U<<_XEN_X86_EMU_LAPIC)
279#define _XEN_X86_EMU_HPET           1
280#define XEN_X86_EMU_HPET            (1U<<_XEN_X86_EMU_HPET)
281#define _XEN_X86_EMU_PM             2
282#define XEN_X86_EMU_PM              (1U<<_XEN_X86_EMU_PM)
283#define _XEN_X86_EMU_RTC            3
284#define XEN_X86_EMU_RTC             (1U<<_XEN_X86_EMU_RTC)
285#define _XEN_X86_EMU_IOAPIC         4
286#define XEN_X86_EMU_IOAPIC          (1U<<_XEN_X86_EMU_IOAPIC)
287#define _XEN_X86_EMU_PIC            5
288#define XEN_X86_EMU_PIC             (1U<<_XEN_X86_EMU_PIC)
289#define _XEN_X86_EMU_VGA            6
290#define XEN_X86_EMU_VGA             (1U<<_XEN_X86_EMU_VGA)
291#define _XEN_X86_EMU_IOMMU          7
292#define XEN_X86_EMU_IOMMU           (1U<<_XEN_X86_EMU_IOMMU)
293#define _XEN_X86_EMU_PIT            8
294#define XEN_X86_EMU_PIT             (1U<<_XEN_X86_EMU_PIT)
295#define _XEN_X86_EMU_USE_PIRQ       9
296#define XEN_X86_EMU_USE_PIRQ        (1U<<_XEN_X86_EMU_USE_PIRQ)
297#define _XEN_X86_EMU_VPCI           10
298#define XEN_X86_EMU_VPCI            (1U<<_XEN_X86_EMU_VPCI)
299
300#define XEN_X86_EMU_ALL             (XEN_X86_EMU_LAPIC | XEN_X86_EMU_HPET |  \
301                                     XEN_X86_EMU_PM | XEN_X86_EMU_RTC |      \
302                                     XEN_X86_EMU_IOAPIC | XEN_X86_EMU_PIC |  \
303                                     XEN_X86_EMU_VGA | XEN_X86_EMU_IOMMU |   \
304                                     XEN_X86_EMU_PIT | XEN_X86_EMU_USE_PIRQ |\
305                                     XEN_X86_EMU_VPCI)
306    uint32_t emulation_flags;
307};
308
309/* Location of online VCPU bitmap. */
310#define XEN_ACPI_CPU_MAP             0xaf00
311#define XEN_ACPI_CPU_MAP_LEN         ((HVM_MAX_VCPUS + 7) / 8)
312
313/* GPE0 bit set during CPU hotplug */
314#define XEN_ACPI_GPE0_CPUHP_BIT      2
315#endif
316
317/*
318 * Representations of architectural CPUID and MSR information.  Used as the
319 * serialised version of Xen's internal representation.
320 */
321typedef struct xen_cpuid_leaf {
322#define XEN_CPUID_NO_SUBLEAF 0xffffffffu
323    uint32_t leaf, subleaf;
324    uint32_t a, b, c, d;
325} xen_cpuid_leaf_t;
326DEFINE_XEN_GUEST_HANDLE(xen_cpuid_leaf_t);
327
328typedef struct xen_msr_entry {
329    uint32_t idx;
330    uint32_t flags; /* Reserved MBZ. */
331    uint64_t val;
332} xen_msr_entry_t;
333DEFINE_XEN_GUEST_HANDLE(xen_msr_entry_t);
334
335#endif /* !__ASSEMBLY__ */
336
337/*
338 * ` enum neg_errnoval
339 * ` HYPERVISOR_fpu_taskswitch(int set);
340 * `
341 * Sets (if set!=0) or clears (if set==0) CR0.TS.
342 */
343
344/*
345 * ` enum neg_errnoval
346 * ` HYPERVISOR_set_debugreg(int regno, unsigned long value);
347 *
348 * ` unsigned long
349 * ` HYPERVISOR_get_debugreg(int regno);
350 * For 0<=reg<=7, returns the debug register value.
351 * For other values of reg, returns ((unsigned long)-EINVAL).
352 * (Unfortunately, this interface is defective.)
353 */
354
355/*
356 * Prefix forces emulation of some non-trapping instructions.
357 * Currently only CPUID.
358 */
359#ifdef __ASSEMBLY__
360#define XEN_EMULATE_PREFIX .byte 0x0f,0x0b,0x78,0x65,0x6e ;
361#define XEN_CPUID          XEN_EMULATE_PREFIX cpuid
362#else
363#define XEN_EMULATE_PREFIX ".byte 0x0f,0x0b,0x78,0x65,0x6e ; "
364#define XEN_CPUID          XEN_EMULATE_PREFIX "cpuid"
365#endif
366
367/*
368 * Debug console IO port, also called "port E9 hack". Each character written
369 * to this IO port will be printed on the hypervisor console, subject to log
370 * level restrictions.
371 */
372#define XEN_HVM_DEBUGCONS_IOPORT 0xe9
373
374#endif /* __XEN_PUBLIC_ARCH_X86_XEN_H__ */
375
376/*
377 * Local variables:
378 * mode: C
379 * c-file-style: "BSD"
380 * c-basic-offset: 4
381 * tab-width: 4
382 * indent-tabs-mode: nil
383 * End:
384 */
385