1  /******************************************************************************
2   * sysctl.h
3   *
4   * System management operations. For use by node control stack.
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) 2002-2006, K Fraser
25   */
26  
27  #ifndef __XEN_PUBLIC_SYSCTL_H__
28  #define __XEN_PUBLIC_SYSCTL_H__
29  
30  #if !defined(__XEN__) && !defined(__XEN_TOOLS__)
31  #error "sysctl operations are intended for use by node control tools only"
32  #endif
33  
34  #include "xen.h"
35  #include "domctl.h"
36  #include "physdev.h"
37  
38  #define XEN_SYSCTL_INTERFACE_VERSION 0x00000013
39  
40  /*
41   * Read console content from Xen buffer ring.
42   */
43  /* XEN_SYSCTL_readconsole */
44  struct xen_sysctl_readconsole {
45      /* IN: Non-zero -> clear after reading. */
46      uint8_t clear;
47      /* IN: Non-zero -> start index specified by @index field. */
48      uint8_t incremental;
49      uint8_t pad0, pad1;
50      /*
51       * IN:  Start index for consuming from ring buffer (if @incremental);
52       * OUT: End index after consuming from ring buffer.
53       */
54      uint32_t index;
55      /* IN: Virtual address to write console data. */
56      XEN_GUEST_HANDLE_64(char) buffer;
57      /* IN: Size of buffer; OUT: Bytes written to buffer. */
58      uint32_t count;
59  };
60  
61  /* Get trace buffers machine base address */
62  /* XEN_SYSCTL_tbuf_op */
63  struct xen_sysctl_tbuf_op {
64      /* IN variables */
65  #define XEN_SYSCTL_TBUFOP_get_info     0
66  #define XEN_SYSCTL_TBUFOP_set_cpu_mask 1
67  #define XEN_SYSCTL_TBUFOP_set_evt_mask 2
68  #define XEN_SYSCTL_TBUFOP_set_size     3
69  #define XEN_SYSCTL_TBUFOP_enable       4
70  #define XEN_SYSCTL_TBUFOP_disable      5
71      uint32_t cmd;
72      /* IN/OUT variables */
73      struct xenctl_bitmap cpu_mask;
74      uint32_t             evt_mask;
75      /* OUT variables */
76      uint64_aligned_t buffer_mfn;
77      uint32_t size;  /* Also an IN variable! */
78  };
79  
80  /*
81   * Get physical information about the host machine
82   */
83  /* XEN_SYSCTL_physinfo */
84   /* The platform supports HVM guests. */
85  #define _XEN_SYSCTL_PHYSCAP_hvm          0
86  #define XEN_SYSCTL_PHYSCAP_hvm           (1u<<_XEN_SYSCTL_PHYSCAP_hvm)
87   /* The platform supports PV guests. */
88  #define _XEN_SYSCTL_PHYSCAP_pv           1
89  #define XEN_SYSCTL_PHYSCAP_pv            (1u<<_XEN_SYSCTL_PHYSCAP_pv)
90   /* The platform supports direct access to I/O devices with IOMMU. */
91  #define _XEN_SYSCTL_PHYSCAP_directio     2
92  #define XEN_SYSCTL_PHYSCAP_directio  (1u<<_XEN_SYSCTL_PHYSCAP_directio)
93  /* The platform supports Hardware Assisted Paging. */
94  #define _XEN_SYSCTL_PHYSCAP_hap          3
95  #define XEN_SYSCTL_PHYSCAP_hap           (1u<<_XEN_SYSCTL_PHYSCAP_hap)
96  /* The platform supports software paging. */
97  #define _XEN_SYSCTL_PHYSCAP_shadow       4
98  #define XEN_SYSCTL_PHYSCAP_shadow        (1u<<_XEN_SYSCTL_PHYSCAP_shadow)
99  /* The platform supports sharing of HAP page tables with the IOMMU. */
100  #define _XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share 5
101  #define XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share  \
102      (1u << _XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share)
103  
104  /* Max XEN_SYSCTL_PHYSCAP_* constant.  Used for ABI checking. */
105  #define XEN_SYSCTL_PHYSCAP_MAX XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share
106  
107  struct xen_sysctl_physinfo {
108      uint32_t threads_per_core;
109      uint32_t cores_per_socket;
110      uint32_t nr_cpus;     /* # CPUs currently online */
111      uint32_t max_cpu_id;  /* Largest possible CPU ID on this host */
112      uint32_t nr_nodes;    /* # nodes currently online */
113      uint32_t max_node_id; /* Largest possible node ID on this host */
114      uint32_t cpu_khz;
115      uint32_t capabilities;/* XEN_SYSCTL_PHYSCAP_??? */
116      uint64_aligned_t total_pages;
117      uint64_aligned_t free_pages;
118      uint64_aligned_t scrub_pages;
119      uint64_aligned_t outstanding_pages;
120      uint64_aligned_t max_mfn; /* Largest possible MFN on this host */
121      uint32_t hw_cap[8];
122  };
123  
124  /*
125   * Get the ID of the current scheduler.
126   */
127  /* XEN_SYSCTL_sched_id */
128  struct xen_sysctl_sched_id {
129      /* OUT variable */
130      uint32_t sched_id;
131  };
132  
133  /* Interface for controlling Xen software performance counters. */
134  /* XEN_SYSCTL_perfc_op */
135  /* Sub-operations: */
136  #define XEN_SYSCTL_PERFCOP_reset 1   /* Reset all counters to zero. */
137  #define XEN_SYSCTL_PERFCOP_query 2   /* Get perfctr information. */
138  struct xen_sysctl_perfc_desc {
139      char         name[80];             /* name of perf counter */
140      uint32_t     nr_vals;              /* number of values for this counter */
141  };
142  typedef struct xen_sysctl_perfc_desc xen_sysctl_perfc_desc_t;
143  DEFINE_XEN_GUEST_HANDLE(xen_sysctl_perfc_desc_t);
144  typedef uint32_t xen_sysctl_perfc_val_t;
145  DEFINE_XEN_GUEST_HANDLE(xen_sysctl_perfc_val_t);
146  
147  struct xen_sysctl_perfc_op {
148      /* IN variables. */
149      uint32_t       cmd;                /*  XEN_SYSCTL_PERFCOP_??? */
150      /* OUT variables. */
151      uint32_t       nr_counters;       /*  number of counters description  */
152      uint32_t       nr_vals;           /*  number of values  */
153      /* counter information (or NULL) */
154      XEN_GUEST_HANDLE_64(xen_sysctl_perfc_desc_t) desc;
155      /* counter values (or NULL) */
156      XEN_GUEST_HANDLE_64(xen_sysctl_perfc_val_t) val;
157  };
158  
159  /* XEN_SYSCTL_getdomaininfolist */
160  struct xen_sysctl_getdomaininfolist {
161      /* IN variables. */
162      domid_t               first_domain;
163      uint32_t              max_domains;
164      XEN_GUEST_HANDLE_64(xen_domctl_getdomaininfo_t) buffer;
165      /* OUT variables. */
166      uint32_t              num_domains;
167  };
168  
169  /* Inject debug keys into Xen. */
170  /* XEN_SYSCTL_debug_keys */
171  struct xen_sysctl_debug_keys {
172      /* IN variables. */
173      XEN_GUEST_HANDLE_64(const_char) keys;
174      uint32_t nr_keys;
175  };
176  
177  /* Get physical CPU information. */
178  /* XEN_SYSCTL_getcpuinfo */
179  struct xen_sysctl_cpuinfo {
180      uint64_aligned_t idletime;
181  };
182  typedef struct xen_sysctl_cpuinfo xen_sysctl_cpuinfo_t;
183  DEFINE_XEN_GUEST_HANDLE(xen_sysctl_cpuinfo_t);
184  struct xen_sysctl_getcpuinfo {
185      /* IN variables. */
186      uint32_t max_cpus;
187      XEN_GUEST_HANDLE_64(xen_sysctl_cpuinfo_t) info;
188      /* OUT variables. */
189      uint32_t nr_cpus;
190  };
191  
192  /* XEN_SYSCTL_availheap */
193  struct xen_sysctl_availheap {
194      /* IN variables. */
195      uint32_t min_bitwidth;  /* Smallest address width (zero if don't care). */
196      uint32_t max_bitwidth;  /* Largest address width (zero if don't care). */
197      int32_t  node;          /* NUMA node of interest (-1 for all nodes). */
198      /* OUT variables. */
199      uint64_aligned_t avail_bytes;/* Bytes available in the specified region. */
200  };
201  
202  /* XEN_SYSCTL_get_pmstat */
203  struct pm_px_val {
204      uint64_aligned_t freq;        /* Px core frequency */
205      uint64_aligned_t residency;   /* Px residency time */
206      uint64_aligned_t count;       /* Px transition count */
207  };
208  typedef struct pm_px_val pm_px_val_t;
209  DEFINE_XEN_GUEST_HANDLE(pm_px_val_t);
210  
211  struct pm_px_stat {
212      uint8_t total;        /* total Px states */
213      uint8_t usable;       /* usable Px states */
214      uint8_t last;         /* last Px state */
215      uint8_t cur;          /* current Px state */
216      XEN_GUEST_HANDLE_64(uint64) trans_pt;   /* Px transition table */
217      XEN_GUEST_HANDLE_64(pm_px_val_t) pt;
218  };
219  
220  struct pm_cx_stat {
221      uint32_t nr;    /* entry nr in triggers & residencies, including C0 */
222      uint32_t last;  /* last Cx state */
223      uint64_aligned_t idle_time;                 /* idle time from boot */
224      XEN_GUEST_HANDLE_64(uint64) triggers;    /* Cx trigger counts */
225      XEN_GUEST_HANDLE_64(uint64) residencies; /* Cx residencies */
226      uint32_t nr_pc;                          /* entry nr in pc[] */
227      uint32_t nr_cc;                          /* entry nr in cc[] */
228      /*
229       * These two arrays may (and generally will) have unused slots; slots not
230       * having a corresponding hardware register will not be written by the
231       * hypervisor. It is therefore up to the caller to put a suitable sentinel
232       * into all slots before invoking the function.
233       * Indexing is 1-biased (PC1/CC1 being at index 0).
234       */
235      XEN_GUEST_HANDLE_64(uint64) pc;
236      XEN_GUEST_HANDLE_64(uint64) cc;
237  };
238  
239  struct xen_sysctl_get_pmstat {
240  #define PMSTAT_CATEGORY_MASK 0xf0
241  #define PMSTAT_PX            0x10
242  #define PMSTAT_CX            0x20
243  #define PMSTAT_get_max_px    (PMSTAT_PX | 0x1)
244  #define PMSTAT_get_pxstat    (PMSTAT_PX | 0x2)
245  #define PMSTAT_reset_pxstat  (PMSTAT_PX | 0x3)
246  #define PMSTAT_get_max_cx    (PMSTAT_CX | 0x1)
247  #define PMSTAT_get_cxstat    (PMSTAT_CX | 0x2)
248  #define PMSTAT_reset_cxstat  (PMSTAT_CX | 0x3)
249      uint32_t type;
250      uint32_t cpuid;
251      union {
252          struct pm_px_stat getpx;
253          struct pm_cx_stat getcx;
254          /* other struct for tx, etc */
255      } u;
256  };
257  
258  /* XEN_SYSCTL_cpu_hotplug */
259  struct xen_sysctl_cpu_hotplug {
260      /* IN variables */
261      uint32_t cpu;   /* Physical cpu. */
262  
263      /* Single CPU enable/disable. */
264  #define XEN_SYSCTL_CPU_HOTPLUG_ONLINE  0
265  #define XEN_SYSCTL_CPU_HOTPLUG_OFFLINE 1
266  
267      /*
268       * SMT enable/disable.
269       *
270       * These two ops loop over all present CPUs, and either online or offline
271       * every non-primary sibling thread (those with a thread id which is not
272       * 0).  This behaviour is chosen to simplify the implementation.
273       *
274       * They are intended as a shorthand for identifying and feeding the cpu
275       * numbers individually to HOTPLUG_{ON,OFF}LINE.
276       *
277       * These are not expected to be used in conjunction with debugging options
278       * such as `maxcpus=` or when other manual configuration of offline cpus
279       * is in use.
280       */
281  #define XEN_SYSCTL_CPU_HOTPLUG_SMT_ENABLE  2
282  #define XEN_SYSCTL_CPU_HOTPLUG_SMT_DISABLE 3
283      uint32_t op;    /* hotplug opcode */
284  };
285  
286  /*
287   * Get/set xen power management, include
288   * 1. cpufreq governors and related parameters
289   */
290  /* XEN_SYSCTL_pm_op */
291  struct xen_userspace {
292      uint32_t scaling_setspeed;
293  };
294  
295  struct xen_ondemand {
296      uint32_t sampling_rate_max;
297      uint32_t sampling_rate_min;
298  
299      uint32_t sampling_rate;
300      uint32_t up_threshold;
301  };
302  
303  /*
304   * cpufreq para name of this structure named
305   * same as sysfs file name of native linux
306   */
307  #define CPUFREQ_NAME_LEN 16
308  struct xen_get_cpufreq_para {
309      /* IN/OUT variable */
310      uint32_t cpu_num;
311      uint32_t freq_num;
312      uint32_t gov_num;
313  
314      /* for all governors */
315      /* OUT variable */
316      XEN_GUEST_HANDLE_64(uint32) affected_cpus;
317      XEN_GUEST_HANDLE_64(uint32) scaling_available_frequencies;
318      XEN_GUEST_HANDLE_64(char)   scaling_available_governors;
319      char scaling_driver[CPUFREQ_NAME_LEN];
320  
321      uint32_t cpuinfo_cur_freq;
322      uint32_t cpuinfo_max_freq;
323      uint32_t cpuinfo_min_freq;
324      uint32_t scaling_cur_freq;
325  
326      char scaling_governor[CPUFREQ_NAME_LEN];
327      uint32_t scaling_max_freq;
328      uint32_t scaling_min_freq;
329  
330      /* for specific governor */
331      union {
332          struct  xen_userspace userspace;
333          struct  xen_ondemand ondemand;
334      } u;
335  
336      int32_t turbo_enabled;
337  };
338  
339  struct xen_set_cpufreq_gov {
340      char scaling_governor[CPUFREQ_NAME_LEN];
341  };
342  
343  struct xen_set_cpufreq_para {
344      #define SCALING_MAX_FREQ           1
345      #define SCALING_MIN_FREQ           2
346      #define SCALING_SETSPEED           3
347      #define SAMPLING_RATE              4
348      #define UP_THRESHOLD               5
349  
350      uint32_t ctrl_type;
351      uint32_t ctrl_value;
352  };
353  
354  struct xen_sysctl_pm_op {
355      #define PM_PARA_CATEGORY_MASK      0xf0
356      #define CPUFREQ_PARA               0x10
357  
358      /* cpufreq command type */
359      #define GET_CPUFREQ_PARA           (CPUFREQ_PARA | 0x01)
360      #define SET_CPUFREQ_GOV            (CPUFREQ_PARA | 0x02)
361      #define SET_CPUFREQ_PARA           (CPUFREQ_PARA | 0x03)
362      #define GET_CPUFREQ_AVGFREQ        (CPUFREQ_PARA | 0x04)
363  
364      /* set/reset scheduler power saving option */
365      #define XEN_SYSCTL_pm_op_set_sched_opt_smt    0x21
366  
367      /*
368       * cpuidle max C-state and max C-sub-state access command:
369       * Set cpuid to 0 for max C-state.
370       * Set cpuid to 1 for max C-sub-state.
371       */
372      #define XEN_SYSCTL_pm_op_get_max_cstate       0x22
373      #define XEN_SYSCTL_pm_op_set_max_cstate       0x23
374  
375      /* set scheduler migration cost value */
376      #define XEN_SYSCTL_pm_op_set_vcpu_migration_delay   0x24
377      #define XEN_SYSCTL_pm_op_get_vcpu_migration_delay   0x25
378  
379      /* enable/disable turbo mode when in dbs governor */
380      #define XEN_SYSCTL_pm_op_enable_turbo               0x26
381      #define XEN_SYSCTL_pm_op_disable_turbo              0x27
382  
383      uint32_t cmd;
384      uint32_t cpuid;
385      union {
386          struct xen_get_cpufreq_para get_para;
387          struct xen_set_cpufreq_gov  set_gov;
388          struct xen_set_cpufreq_para set_para;
389          uint64_aligned_t get_avgfreq;
390          uint32_t                    set_sched_opt_smt;
391  #define XEN_SYSCTL_CX_UNLIMITED 0xffffffff
392          uint32_t                    get_max_cstate;
393          uint32_t                    set_max_cstate;
394      } u;
395  };
396  
397  /* XEN_SYSCTL_page_offline_op */
398  struct xen_sysctl_page_offline_op {
399      /* IN: range of page to be offlined */
400  #define sysctl_page_offline     1
401  #define sysctl_page_online      2
402  #define sysctl_query_page_offline  3
403      uint32_t cmd;
404      uint32_t start;
405      uint32_t end;
406      /* OUT: result of page offline request */
407      /*
408       * bit 0~15: result flags
409       * bit 16~31: owner
410       */
411      XEN_GUEST_HANDLE(uint32) status;
412  };
413  
414  #define PG_OFFLINE_STATUS_MASK    (0xFFUL)
415  
416  /* The result is invalid, i.e. HV does not handle it */
417  #define PG_OFFLINE_INVALID   (0x1UL << 0)
418  
419  #define PG_OFFLINE_OFFLINED  (0x1UL << 1)
420  #define PG_OFFLINE_PENDING   (0x1UL << 2)
421  #define PG_OFFLINE_FAILED    (0x1UL << 3)
422  #define PG_OFFLINE_AGAIN     (0x1UL << 4)
423  
424  #define PG_ONLINE_FAILED     PG_OFFLINE_FAILED
425  #define PG_ONLINE_ONLINED    PG_OFFLINE_OFFLINED
426  
427  #define PG_OFFLINE_STATUS_OFFLINED              (0x1UL << 1)
428  #define PG_OFFLINE_STATUS_ONLINE                (0x1UL << 2)
429  #define PG_OFFLINE_STATUS_OFFLINE_PENDING       (0x1UL << 3)
430  #define PG_OFFLINE_STATUS_BROKEN                (0x1UL << 4)
431  
432  #define PG_OFFLINE_MISC_MASK    (0xFFUL << 4)
433  
434  /* valid when PG_OFFLINE_FAILED or PG_OFFLINE_PENDING */
435  #define PG_OFFLINE_XENPAGE   (0x1UL << 8)
436  #define PG_OFFLINE_DOM0PAGE  (0x1UL << 9)
437  #define PG_OFFLINE_ANONYMOUS (0x1UL << 10)
438  #define PG_OFFLINE_NOT_CONV_RAM   (0x1UL << 11)
439  #define PG_OFFLINE_OWNED     (0x1UL << 12)
440  
441  #define PG_OFFLINE_BROKEN    (0x1UL << 13)
442  #define PG_ONLINE_BROKEN     PG_OFFLINE_BROKEN
443  
444  #define PG_OFFLINE_OWNER_SHIFT 16
445  
446  /* XEN_SYSCTL_lockprof_op */
447  /* Sub-operations: */
448  #define XEN_SYSCTL_LOCKPROF_reset 1   /* Reset all profile data to zero. */
449  #define XEN_SYSCTL_LOCKPROF_query 2   /* Get lock profile information. */
450  /* Record-type: */
451  #define LOCKPROF_TYPE_GLOBAL      0   /* global lock, idx meaningless */
452  #define LOCKPROF_TYPE_PERDOM      1   /* per-domain lock, idx is domid */
453  #define LOCKPROF_TYPE_N           2   /* number of types */
454  struct xen_sysctl_lockprof_data {
455      char     name[40];     /* lock name (may include up to 2 %d specifiers) */
456      int32_t  type;         /* LOCKPROF_TYPE_??? */
457      int32_t  idx;          /* index (e.g. domain id) */
458      uint64_aligned_t lock_cnt;     /* # of locking succeeded */
459      uint64_aligned_t block_cnt;    /* # of wait for lock */
460      uint64_aligned_t lock_time;    /* nsecs lock held */
461      uint64_aligned_t block_time;   /* nsecs waited for lock */
462  };
463  typedef struct xen_sysctl_lockprof_data xen_sysctl_lockprof_data_t;
464  DEFINE_XEN_GUEST_HANDLE(xen_sysctl_lockprof_data_t);
465  struct xen_sysctl_lockprof_op {
466      /* IN variables. */
467      uint32_t       cmd;               /* XEN_SYSCTL_LOCKPROF_??? */
468      uint32_t       max_elem;          /* size of output buffer */
469      /* OUT variables (query only). */
470      uint32_t       nr_elem;           /* number of elements available */
471      uint64_aligned_t time;            /* nsecs of profile measurement */
472      /* profile information (or NULL) */
473      XEN_GUEST_HANDLE_64(xen_sysctl_lockprof_data_t) data;
474  };
475  
476  /* XEN_SYSCTL_cputopoinfo */
477  #define XEN_INVALID_CORE_ID     (~0U)
478  #define XEN_INVALID_SOCKET_ID   (~0U)
479  #define XEN_INVALID_NODE_ID     (~0U)
480  
481  struct xen_sysctl_cputopo {
482      uint32_t core;
483      uint32_t socket;
484      uint32_t node;
485  };
486  typedef struct xen_sysctl_cputopo xen_sysctl_cputopo_t;
487  DEFINE_XEN_GUEST_HANDLE(xen_sysctl_cputopo_t);
488  
489  /*
490   * IN:
491   *  - a NULL 'cputopo' handle is a request for maximun 'num_cpus'.
492   *  - otherwise it's the number of entries in 'cputopo'
493   *
494   * OUT:
495   *  - If 'num_cpus' is less than the number Xen wants to write but the handle
496   *    handle is not a NULL one, partial data gets returned and 'num_cpus' gets
497   *    updated to reflect the intended number.
498   *  - Otherwise, 'num_cpus' shall indicate the number of entries written, which
499   *    may be less than the input value.
500   */
501  struct xen_sysctl_cputopoinfo {
502      uint32_t num_cpus;
503      XEN_GUEST_HANDLE_64(xen_sysctl_cputopo_t) cputopo;
504  };
505  
506  /* XEN_SYSCTL_numainfo */
507  #define XEN_INVALID_MEM_SZ     (~0U)
508  #define XEN_INVALID_NODE_DIST  (~0U)
509  
510  struct xen_sysctl_meminfo {
511      uint64_t memsize;
512      uint64_t memfree;
513  };
514  typedef struct xen_sysctl_meminfo xen_sysctl_meminfo_t;
515  DEFINE_XEN_GUEST_HANDLE(xen_sysctl_meminfo_t);
516  
517  /*
518   * IN:
519   *  - Both 'meminfo' and 'distance' handles being null is a request
520   *    for maximum value of 'num_nodes'.
521   *  - Otherwise it's the number of entries in 'meminfo' and square root
522   *    of number of entries in 'distance' (when corresponding handle is
523   *    non-null)
524   *
525   * OUT:
526   *  - If 'num_nodes' is less than the number Xen wants to write but either
527   *    handle is not a NULL one, partial data gets returned and 'num_nodes'
528   *    gets updated to reflect the intended number.
529   *  - Otherwise, 'num_nodes' shall indicate the number of entries written, which
530   *    may be less than the input value.
531   */
532  
533  struct xen_sysctl_numainfo {
534      uint32_t num_nodes;
535  
536      XEN_GUEST_HANDLE_64(xen_sysctl_meminfo_t) meminfo;
537  
538      /*
539       * Distance between nodes 'i' and 'j' is stored in index 'i*N + j',
540       * where N is the number of nodes that will be returned in 'num_nodes'
541       * (i.e. not 'num_nodes' provided by the caller)
542       */
543      XEN_GUEST_HANDLE_64(uint32) distance;
544  };
545  
546  /* XEN_SYSCTL_cpupool_op */
547  #define XEN_SYSCTL_CPUPOOL_OP_CREATE                1  /* C */
548  #define XEN_SYSCTL_CPUPOOL_OP_DESTROY               2  /* D */
549  #define XEN_SYSCTL_CPUPOOL_OP_INFO                  3  /* I */
550  #define XEN_SYSCTL_CPUPOOL_OP_ADDCPU                4  /* A */
551  #define XEN_SYSCTL_CPUPOOL_OP_RMCPU                 5  /* R */
552  #define XEN_SYSCTL_CPUPOOL_OP_MOVEDOMAIN            6  /* M */
553  #define XEN_SYSCTL_CPUPOOL_OP_FREEINFO              7  /* F */
554  #define XEN_SYSCTL_CPUPOOL_PAR_ANY     0xFFFFFFFF
555  struct xen_sysctl_cpupool_op {
556      uint32_t op;          /* IN */
557      uint32_t cpupool_id;  /* IN: CDIARM OUT: CI */
558      uint32_t sched_id;    /* IN: C      OUT: I  */
559      uint32_t domid;       /* IN: M              */
560      uint32_t cpu;         /* IN: AR             */
561      uint32_t n_dom;       /*            OUT: I  */
562      struct xenctl_bitmap cpumap; /*     OUT: IF */
563  };
564  
565  /*
566   * Error return values of cpupool operations:
567   *
568   * -EADDRINUSE:
569   *  XEN_SYSCTL_CPUPOOL_OP_RMCPU: A vcpu is temporarily pinned to the cpu
570   *    which is to be removed from a cpupool.
571   * -EADDRNOTAVAIL:
572   *  XEN_SYSCTL_CPUPOOL_OP_ADDCPU, XEN_SYSCTL_CPUPOOL_OP_RMCPU: A previous
573   *    request to remove a cpu from a cpupool was terminated with -EAGAIN
574   *    and has not been retried using the same parameters.
575   * -EAGAIN:
576   *  XEN_SYSCTL_CPUPOOL_OP_RMCPU: The cpu can't be removed from the cpupool
577   *    as it is active in the hypervisor. A retry will succeed soon.
578   * -EBUSY:
579   *  XEN_SYSCTL_CPUPOOL_OP_DESTROY, XEN_SYSCTL_CPUPOOL_OP_RMCPU: A cpupool
580   *    can't be destroyed or the last cpu can't be removed as there is still
581   *    a running domain in that cpupool.
582   * -EEXIST:
583   *  XEN_SYSCTL_CPUPOOL_OP_CREATE: A cpupool_id was specified and is already
584   *    existing.
585   * -EINVAL:
586   *  XEN_SYSCTL_CPUPOOL_OP_ADDCPU, XEN_SYSCTL_CPUPOOL_OP_RMCPU: An illegal
587   *    cpu was specified (cpu does not exist).
588   *  XEN_SYSCTL_CPUPOOL_OP_MOVEDOMAIN: An illegal domain was specified
589   *    (domain id illegal or not suitable for operation).
590   * -ENODEV:
591   *  XEN_SYSCTL_CPUPOOL_OP_ADDCPU, XEN_SYSCTL_CPUPOOL_OP_RMCPU: The specified
592   *    cpu is either not free (add) or not member of the specified cpupool
593   *    (remove).
594   * -ENOENT:
595   *  all: The cpupool with the specified cpupool_id doesn't exist.
596   *
597   * Some common error return values like -ENOMEM and -EFAULT are possible for
598   * all the operations.
599   */
600  
601  #define ARINC653_MAX_DOMAINS_PER_SCHEDULE   64
602  /*
603   * This structure is used to pass a new ARINC653 schedule from a
604   * privileged domain (ie dom0) to Xen.
605   */
606  struct xen_sysctl_arinc653_schedule {
607      /* major_frame holds the time for the new schedule's major frame
608       * in nanoseconds. */
609      uint64_aligned_t     major_frame;
610      /* num_sched_entries holds how many of the entries in the
611       * sched_entries[] array are valid. */
612      uint8_t     num_sched_entries;
613      /* The sched_entries array holds the actual schedule entries. */
614      struct {
615          /* dom_handle must match a domain's UUID */
616          xen_domain_handle_t dom_handle;
617          /* If a domain has multiple VCPUs, vcpu_id specifies which one
618           * this schedule entry applies to. It should be set to 0 if
619           * there is only one VCPU for the domain. */
620          unsigned int vcpu_id;
621          /* runtime specifies the amount of time that should be allocated
622           * to this VCPU per major frame. It is specified in nanoseconds */
623          uint64_aligned_t runtime;
624      } sched_entries[ARINC653_MAX_DOMAINS_PER_SCHEDULE];
625  };
626  typedef struct xen_sysctl_arinc653_schedule xen_sysctl_arinc653_schedule_t;
627  DEFINE_XEN_GUEST_HANDLE(xen_sysctl_arinc653_schedule_t);
628  
629  /*
630   * Valid range for context switch rate limit (in microseconds).
631   * Applicable to Credit and Credit2 schedulers.
632   */
633  #define XEN_SYSCTL_SCHED_RATELIMIT_MAX 500000
634  #define XEN_SYSCTL_SCHED_RATELIMIT_MIN 100
635  
636  struct xen_sysctl_credit_schedule {
637      /* Length of timeslice in milliseconds */
638  #define XEN_SYSCTL_CSCHED_TSLICE_MAX 1000
639  #define XEN_SYSCTL_CSCHED_TSLICE_MIN 1
640      unsigned tslice_ms;
641      unsigned ratelimit_us;
642      /*
643       * How long we consider a vCPU to be cache-hot on the
644       * CPU where it has run (max 100ms, in microseconds)
645      */
646  #define XEN_SYSCTL_CSCHED_MGR_DLY_MAX_US (100 * 1000)
647      unsigned vcpu_migr_delay_us;
648  };
649  
650  struct xen_sysctl_credit2_schedule {
651      unsigned ratelimit_us;
652  };
653  
654  /* XEN_SYSCTL_scheduler_op */
655  /* Set or get info? */
656  #define XEN_SYSCTL_SCHEDOP_putinfo 0
657  #define XEN_SYSCTL_SCHEDOP_getinfo 1
658  struct xen_sysctl_scheduler_op {
659      uint32_t cpupool_id; /* Cpupool whose scheduler is to be targetted. */
660      uint32_t sched_id;   /* XEN_SCHEDULER_* (domctl.h) */
661      uint32_t cmd;        /* XEN_SYSCTL_SCHEDOP_* */
662      union {
663          struct xen_sysctl_sched_arinc653 {
664              XEN_GUEST_HANDLE_64(xen_sysctl_arinc653_schedule_t) schedule;
665          } sched_arinc653;
666          struct xen_sysctl_credit_schedule sched_credit;
667          struct xen_sysctl_credit2_schedule sched_credit2;
668      } u;
669  };
670  
671  /*
672   * Output format of gcov data:
673   *
674   * XEN_GCOV_FORMAT_MAGIC XEN_GCOV_RECORD ... XEN_GCOV_RECORD
675   *
676   * That is, one magic number followed by 0 or more record.
677   *
678   * The magic number is stored as an uint32_t field.
679   *
680   * The record is packed and variable in length. It has the form:
681   *
682   *  filename: a NULL terminated path name extracted from gcov, used to
683   *            create the name of gcda file.
684   *  size:     a uint32_t field indicating the size of the payload, the
685   *            unit is byte.
686   *  payload:  the actual payload, length is `size' bytes.
687   *
688   * Userspace tool will split the record to different files.
689   */
690  
691  #define XEN_GCOV_FORMAT_MAGIC    0x58434f56 /* XCOV */
692  
693  /*
694   * Ouput format of LLVM coverage data is just a raw stream, as would be
695   * written by the compiler_rt run time library into a .profraw file. There
696   * are no special Xen tags or delimiters because none are needed.
697   */
698  
699  #define XEN_SYSCTL_COVERAGE_get_size 0 /* Get total size of output data */
700  #define XEN_SYSCTL_COVERAGE_read     1 /* Read output data */
701  #define XEN_SYSCTL_COVERAGE_reset    2 /* Reset all counters */
702  
703  struct xen_sysctl_coverage_op {
704      uint32_t cmd;
705      uint32_t size; /* IN/OUT: size of the buffer  */
706      XEN_GUEST_HANDLE_64(char) buffer; /* OUT */
707  };
708  
709  #define XEN_SYSCTL_PSR_CMT_get_total_rmid            0
710  #define XEN_SYSCTL_PSR_CMT_get_l3_upscaling_factor   1
711  /* The L3 cache size is returned in KB unit */
712  #define XEN_SYSCTL_PSR_CMT_get_l3_cache_size         2
713  #define XEN_SYSCTL_PSR_CMT_enabled                   3
714  #define XEN_SYSCTL_PSR_CMT_get_l3_event_mask         4
715  struct xen_sysctl_psr_cmt_op {
716      uint32_t cmd;       /* IN: XEN_SYSCTL_PSR_CMT_* */
717      uint32_t flags;     /* padding variable, may be extended for future use */
718      union {
719          uint64_t data;  /* OUT */
720          struct {
721              uint32_t cpu;   /* IN */
722              uint32_t rsvd;
723          } l3_cache;
724      } u;
725  };
726  
727  /* XEN_SYSCTL_pcitopoinfo */
728  #define XEN_INVALID_DEV (XEN_INVALID_NODE_ID - 1)
729  struct xen_sysctl_pcitopoinfo {
730      /*
731       * IN: Number of elements in 'devs' and 'nodes' arrays.
732       * OUT: Number of processed elements of those arrays.
733       */
734      uint32_t num_devs;
735  
736      /* IN: list of devices for which node IDs are requested. */
737      XEN_GUEST_HANDLE_64(physdev_pci_device_t) devs;
738  
739      /*
740       * OUT: node identifier for each device.
741       * If information for a particular device is not available then
742       * corresponding entry will be set to XEN_INVALID_NODE_ID. If
743       * device is not known to the hypervisor then XEN_INVALID_DEV
744       * will be provided.
745       */
746      XEN_GUEST_HANDLE_64(uint32) nodes;
747  };
748  
749  #define XEN_SYSCTL_PSR_get_l3_info               0
750  #define XEN_SYSCTL_PSR_get_l2_info               1
751  #define XEN_SYSCTL_PSR_get_mba_info              2
752  struct xen_sysctl_psr_alloc {
753      uint32_t cmd;       /* IN: XEN_SYSCTL_PSR_* */
754      uint32_t target;    /* IN */
755      union {
756          struct {
757              uint32_t cbm_len;   /* OUT: CBM length */
758              uint32_t cos_max;   /* OUT: Maximum COS */
759  #define XEN_SYSCTL_PSR_CAT_L3_CDP       (1u << 0)
760              uint32_t flags;     /* OUT: CAT flags */
761          } cat_info;
762  
763          struct {
764              uint32_t thrtl_max; /* OUT: Maximum throttle */
765              uint32_t cos_max;   /* OUT: Maximum COS */
766  #define XEN_SYSCTL_PSR_MBA_LINEAR      (1u << 0)
767              uint32_t flags;     /* OUT: MBA flags */
768          } mba_info;
769      } u;
770  };
771  
772  /*
773   * XEN_SYSCTL_get_cpu_levelling_caps (x86 specific)
774   *
775   * Return hardware capabilities concerning masking or faulting of the cpuid
776   * instruction for PV guests.
777   */
778  struct xen_sysctl_cpu_levelling_caps {
779  #define XEN_SYSCTL_CPU_LEVELCAP_faulting    (1ul <<  0) /* CPUID faulting    */
780  #define XEN_SYSCTL_CPU_LEVELCAP_ecx         (1ul <<  1) /* 0x00000001.ecx    */
781  #define XEN_SYSCTL_CPU_LEVELCAP_edx         (1ul <<  2) /* 0x00000001.edx    */
782  #define XEN_SYSCTL_CPU_LEVELCAP_extd_ecx    (1ul <<  3) /* 0x80000001.ecx    */
783  #define XEN_SYSCTL_CPU_LEVELCAP_extd_edx    (1ul <<  4) /* 0x80000001.edx    */
784  #define XEN_SYSCTL_CPU_LEVELCAP_xsave_eax   (1ul <<  5) /* 0x0000000D:1.eax  */
785  #define XEN_SYSCTL_CPU_LEVELCAP_thermal_ecx (1ul <<  6) /* 0x00000006.ecx    */
786  #define XEN_SYSCTL_CPU_LEVELCAP_l7s0_eax    (1ul <<  7) /* 0x00000007:0.eax  */
787  #define XEN_SYSCTL_CPU_LEVELCAP_l7s0_ebx    (1ul <<  8) /* 0x00000007:0.ebx  */
788      uint32_t caps;
789  };
790  
791  /*
792   * XEN_SYSCTL_get_cpu_featureset (x86 specific)
793   *
794   * Return information about featuresets available on this host.
795   *  -  Raw: The real cpuid values.
796   *  - Host: The values Xen is using, (after command line overrides, etc).
797   *  -   PV: Maximum set of features which can be given to a PV guest.
798   *  -  HVM: Maximum set of features which can be given to a HVM guest.
799   * May fail with -EOPNOTSUPP if querying for PV or HVM data when support is
800   * compiled out of Xen.
801   */
802  struct xen_sysctl_cpu_featureset {
803  #define XEN_SYSCTL_cpu_featureset_raw      0
804  #define XEN_SYSCTL_cpu_featureset_host     1
805  #define XEN_SYSCTL_cpu_featureset_pv       2
806  #define XEN_SYSCTL_cpu_featureset_hvm      3
807      uint32_t index;       /* IN: Which featureset to query? */
808      uint32_t nr_features; /* IN/OUT: Number of entries in/written to
809                             * 'features', or the maximum number of features if
810                             * the guest handle is NULL.  NB. All featuresets
811                             * come from the same numberspace, so have the same
812                             * maximum length. */
813      XEN_GUEST_HANDLE_64(uint32) features; /* OUT: */
814  };
815  
816  /*
817   * XEN_SYSCTL_LIVEPATCH_op
818   *
819   * Refer to the docs/unstable/misc/livepatch.markdown
820   * for the design details of this hypercall.
821   *
822   * There are four sub-ops:
823   *  XEN_SYSCTL_LIVEPATCH_UPLOAD (0)
824   *  XEN_SYSCTL_LIVEPATCH_GET (1)
825   *  XEN_SYSCTL_LIVEPATCH_LIST (2)
826   *  XEN_SYSCTL_LIVEPATCH_ACTION (3)
827   *
828   * The normal sequence of sub-ops is to:
829   *  1) XEN_SYSCTL_LIVEPATCH_UPLOAD to upload the payload. If errors STOP.
830   *  2) XEN_SYSCTL_LIVEPATCH_GET to check the `->rc`. If -XEN_EAGAIN spin.
831   *     If zero go to next step.
832   *  3) XEN_SYSCTL_LIVEPATCH_ACTION with LIVEPATCH_ACTION_APPLY to apply the patch.
833   *  4) XEN_SYSCTL_LIVEPATCH_GET to check the `->rc`. If in -XEN_EAGAIN spin.
834   *     If zero exit with success.
835   */
836  
837  #define LIVEPATCH_PAYLOAD_VERSION 2
838  /*
839   * .livepatch.funcs structure layout defined in the `Payload format`
840   * section in the Live Patch design document.
841   *
842   * We guard this with __XEN__ as toolstacks SHOULD not use it.
843   */
844  #ifdef __XEN__
845  #define LIVEPATCH_OPAQUE_SIZE 31
846  
847  struct livepatch_expectation {
848      uint8_t enabled : 1;
849      uint8_t len : 5;        /* Length of data up to LIVEPATCH_OPAQUE_SIZE
850                                 (5 bits is enough for now) */
851      uint8_t rsv : 2;        /* Reserved. Zero value */
852      uint8_t data[LIVEPATCH_OPAQUE_SIZE]; /* Same size as opaque[] buffer of
853                                              struct livepatch_func. This is the
854                                              max number of bytes to be patched */
855  };
856  typedef struct livepatch_expectation livepatch_expectation_t;
857  
858  typedef enum livepatch_func_state {
859      LIVEPATCH_FUNC_NOT_APPLIED,
860      LIVEPATCH_FUNC_APPLIED
861  } livepatch_func_state_t;
862  
863  struct livepatch_func {
864      const char *name;       /* Name of function to be patched. */
865      void *new_addr;
866      void *old_addr;
867      uint32_t new_size;
868      uint32_t old_size;
869      uint8_t version;        /* MUST be LIVEPATCH_PAYLOAD_VERSION. */
870      uint8_t opaque[LIVEPATCH_OPAQUE_SIZE];
871      uint8_t applied;
872      uint8_t _pad[7];
873      livepatch_expectation_t expect;
874  };
875  typedef struct livepatch_func livepatch_func_t;
876  #endif
877  
878  /*
879   * Structure describing an ELF payload. Uniquely identifies the
880   * payload. Should be human readable.
881   * Recommended length is upto XEN_LIVEPATCH_NAME_SIZE.
882   * Includes the NUL terminator.
883   */
884  #define XEN_LIVEPATCH_NAME_SIZE 128
885  struct xen_livepatch_name {
886      XEN_GUEST_HANDLE_64(char) name;         /* IN: pointer to name. */
887      uint16_t size;                          /* IN: size of name. May be upto
888                                                 XEN_LIVEPATCH_NAME_SIZE. */
889      uint16_t pad[3];                        /* IN: MUST be zero. */
890  };
891  
892  /*
893   * Upload a payload to the hypervisor. The payload is verified
894   * against basic checks and if there are any issues the proper return code
895   * will be returned. The payload is not applied at this time - that is
896   * controlled by XEN_SYSCTL_LIVEPATCH_ACTION.
897   *
898   * The return value is zero if the payload was succesfully uploaded.
899   * Otherwise an EXX return value is provided. Duplicate `name` are not
900   * supported.
901   *
902   * The payload at this point is verified against basic checks.
903   *
904   * The `payload` is the ELF payload as mentioned in the `Payload format`
905   * section in the Live Patch design document.
906   */
907  #define XEN_SYSCTL_LIVEPATCH_UPLOAD 0
908  struct xen_sysctl_livepatch_upload {
909      struct xen_livepatch_name name;         /* IN, name of the patch. */
910      uint64_t size;                          /* IN, size of the ELF file. */
911      XEN_GUEST_HANDLE_64(uint8) payload;     /* IN, the ELF file. */
912  };
913  
914  /*
915   * Retrieve an status of an specific payload.
916   *
917   * Upon completion the `struct xen_livepatch_status` is updated.
918   *
919   * The return value is zero on success and XEN_EXX on failure. This operation
920   * is synchronous and does not require preemption.
921   */
922  #define XEN_SYSCTL_LIVEPATCH_GET 1
923  
924  struct xen_livepatch_status {
925  #define LIVEPATCH_STATE_CHECKED      1
926  #define LIVEPATCH_STATE_APPLIED      2
927      uint32_t state;                /* OUT: LIVEPATCH_STATE_*. */
928      int32_t rc;                    /* OUT: 0 if no error, otherwise -XEN_EXX. */
929  };
930  typedef struct xen_livepatch_status xen_livepatch_status_t;
931  DEFINE_XEN_GUEST_HANDLE(xen_livepatch_status_t);
932  
933  struct xen_sysctl_livepatch_get {
934      struct xen_livepatch_name name;         /* IN, name of the payload. */
935      struct xen_livepatch_status status;     /* IN/OUT, state of it. */
936  };
937  
938  /*
939   * Retrieve an array of abbreviated status, names and metadata of payloads that
940   * are loaded in the hypervisor.
941   *
942   * If the hypercall returns an positive number, it is the number (up to `nr`)
943   * of the payloads returned, along with `nr` updated with the number of remaining
944   * payloads, `version` updated (it may be the same across hypercalls. If it varies
945   * the data is stale and further calls could fail), `name_total_size` and
946   * `metadata_total_size` containing total sizes of transferred data for both the
947   * arrays.
948   * The `status`, `name`, `len`, `metadata` and `metadata_len` are updated at their
949   * designed index value (`idx`) with the returned value of data.
950   *
951   * If the hypercall returns E2BIG the `nr` is too big and should be
952   * lowered. The upper limit of `nr` is left to the implemention.
953   *
954   * Note that due to the asynchronous nature of hypercalls the domain might have
955   * added or removed the number of payloads making this information stale. It is
956   * the responsibility of the toolstack to use the `version` field to check
957   * between each invocation. if the version differs it should discard the stale
958   * data and start from scratch. It is OK for the toolstack to use the new
959   * `version` field.
960   */
961  #define XEN_SYSCTL_LIVEPATCH_LIST 2
962  struct xen_sysctl_livepatch_list {
963      uint32_t version;                       /* OUT: Hypervisor stamps value.
964                                                 If varies between calls, we are
965                                               * getting stale data. */
966      uint32_t idx;                           /* IN: Index into hypervisor list. */
967      uint32_t nr;                            /* IN: How many status, name, and len
968                                                 should fill out. Can be zero to get
969                                                 amount of payloads and version.
970                                                 OUT: How many payloads left. */
971      uint32_t pad;                           /* IN: Must be zero. */
972      uint32_t name_total_size;               /* OUT: Total size of all transfer names */
973      uint32_t metadata_total_size;           /* OUT: Total size of all transfer metadata */
974      XEN_GUEST_HANDLE_64(xen_livepatch_status_t) status;  /* OUT. Must have enough
975                                                 space allocate for nr of them. */
976      XEN_GUEST_HANDLE_64(char) name;         /* OUT: Array of names. Each member
977                                                 may have an arbitrary length up to
978                                                 XEN_LIVEPATCH_NAME_SIZE bytes. Must have
979                                                 nr of them. */
980      XEN_GUEST_HANDLE_64(uint32) len;        /* OUT: Array of lengths of name's.
981                                                 Must have nr of them. */
982      XEN_GUEST_HANDLE_64(char) metadata;     /* OUT: Array of metadata strings. Each
983                                                 member may have an arbitrary length.
984                                                 Must have nr of them. */
985      XEN_GUEST_HANDLE_64(uint32) metadata_len;  /* OUT: Array of lengths of metadata's.
986                                                    Must have nr of them. */
987  };
988  
989  /*
990   * Perform an operation on the payload structure referenced by the `name` field.
991   * The operation request is asynchronous and the status should be retrieved
992   * by using either XEN_SYSCTL_LIVEPATCH_GET or XEN_SYSCTL_LIVEPATCH_LIST hypercall.
993   */
994  #define XEN_SYSCTL_LIVEPATCH_ACTION 3
995  struct xen_sysctl_livepatch_action {
996      struct xen_livepatch_name name;         /* IN, name of the patch. */
997  #define LIVEPATCH_ACTION_UNLOAD       1
998  #define LIVEPATCH_ACTION_REVERT       2
999  #define LIVEPATCH_ACTION_APPLY        3
1000  #define LIVEPATCH_ACTION_REPLACE      4
1001      uint32_t cmd;                           /* IN: LIVEPATCH_ACTION_*. */
1002      uint32_t timeout;                       /* IN: If zero then uses */
1003                                              /* hypervisor default. */
1004                                              /* Or upper bound of time (ns) */
1005                                              /* for operation to take. */
1006  
1007  /*
1008   * Override default inter-module buildid dependency chain enforcement.
1009   * Check only if module is built for given hypervisor by comparing buildid.
1010   */
1011  #define LIVEPATCH_ACTION_APPLY_NODEPS (1 << 0)
1012      uint32_t flags;                         /* IN: action flags. */
1013                                              /* Provide additional parameters */
1014                                              /* for an action. */
1015      uint32_t pad;                           /* IN: Always zero. */
1016  };
1017  
1018  struct xen_sysctl_livepatch_op {
1019      uint32_t cmd;                           /* IN: XEN_SYSCTL_LIVEPATCH_*. */
1020      uint32_t pad;                           /* IN: Always zero. */
1021      union {
1022          struct xen_sysctl_livepatch_upload upload;
1023          struct xen_sysctl_livepatch_list list;
1024          struct xen_sysctl_livepatch_get get;
1025          struct xen_sysctl_livepatch_action action;
1026      } u;
1027  };
1028  
1029  #if defined(__i386__) || defined(__x86_64__)
1030  /*
1031   * XEN_SYSCTL_get_cpu_policy (x86 specific)
1032   *
1033   * Return information about CPUID and MSR policies available on this host.
1034   *  -       Raw: The real H/W values.
1035   *  -      Host: The values Xen is using, (after command line overrides, etc).
1036   *  -     Max_*: Maximum set of features a PV or HVM guest can use.  Includes
1037   *               experimental features outside of security support.
1038   *  - Default_*: Default set of features a PV or HVM guest can use.  This is
1039   *               the security supported set.
1040   * May fail with -EOPNOTSUPP if querying for PV or HVM data when support is
1041   * compiled out of Xen.
1042   */
1043  struct xen_sysctl_cpu_policy {
1044  #define XEN_SYSCTL_cpu_policy_raw          0
1045  #define XEN_SYSCTL_cpu_policy_host         1
1046  #define XEN_SYSCTL_cpu_policy_pv_max       2
1047  #define XEN_SYSCTL_cpu_policy_hvm_max      3
1048  #define XEN_SYSCTL_cpu_policy_pv_default   4
1049  #define XEN_SYSCTL_cpu_policy_hvm_default  5
1050      uint32_t index;       /* IN: Which policy to query? */
1051      uint32_t nr_leaves;   /* IN/OUT: Number of leaves in/written to
1052                             * 'cpuid_policy', or the maximum number of leaves
1053                             * if the guest handle is NULL. */
1054      uint32_t nr_msrs;     /* IN/OUT: Number of MSRs in/written to
1055                             * 'msr_policy', or the maximum number of MSRs if
1056                             * the guest handle is NULL. */
1057      uint32_t _rsvd;       /* Must be zero. */
1058      XEN_GUEST_HANDLE_64(xen_cpuid_leaf_t) cpuid_policy; /* OUT */
1059      XEN_GUEST_HANDLE_64(xen_msr_entry_t) msr_policy;    /* OUT */
1060  };
1061  typedef struct xen_sysctl_cpu_policy xen_sysctl_cpu_policy_t;
1062  DEFINE_XEN_GUEST_HANDLE(xen_sysctl_cpu_policy_t);
1063  #endif
1064  
1065  struct xen_sysctl {
1066      uint32_t cmd;
1067  #define XEN_SYSCTL_readconsole                    1
1068  #define XEN_SYSCTL_tbuf_op                        2
1069  #define XEN_SYSCTL_physinfo                       3
1070  #define XEN_SYSCTL_sched_id                       4
1071  #define XEN_SYSCTL_perfc_op                       5
1072  #define XEN_SYSCTL_getdomaininfolist              6
1073  #define XEN_SYSCTL_debug_keys                     7
1074  #define XEN_SYSCTL_getcpuinfo                     8
1075  #define XEN_SYSCTL_availheap                      9
1076  #define XEN_SYSCTL_get_pmstat                    10
1077  #define XEN_SYSCTL_cpu_hotplug                   11
1078  #define XEN_SYSCTL_pm_op                         12
1079  #define XEN_SYSCTL_page_offline_op               14
1080  #define XEN_SYSCTL_lockprof_op                   15
1081  #define XEN_SYSCTL_cputopoinfo                   16
1082  #define XEN_SYSCTL_numainfo                      17
1083  #define XEN_SYSCTL_cpupool_op                    18
1084  #define XEN_SYSCTL_scheduler_op                  19
1085  #define XEN_SYSCTL_coverage_op                   20
1086  #define XEN_SYSCTL_psr_cmt_op                    21
1087  #define XEN_SYSCTL_pcitopoinfo                   22
1088  #define XEN_SYSCTL_psr_alloc                     23
1089  /* #define XEN_SYSCTL_tmem_op                       24 */
1090  #define XEN_SYSCTL_get_cpu_levelling_caps        25
1091  #define XEN_SYSCTL_get_cpu_featureset            26
1092  #define XEN_SYSCTL_livepatch_op                  27
1093  /* #define XEN_SYSCTL_set_parameter              28 */
1094  #define XEN_SYSCTL_get_cpu_policy                29
1095      uint32_t interface_version; /* XEN_SYSCTL_INTERFACE_VERSION */
1096      union {
1097          struct xen_sysctl_readconsole       readconsole;
1098          struct xen_sysctl_tbuf_op           tbuf_op;
1099          struct xen_sysctl_physinfo          physinfo;
1100          struct xen_sysctl_cputopoinfo       cputopoinfo;
1101          struct xen_sysctl_pcitopoinfo       pcitopoinfo;
1102          struct xen_sysctl_numainfo          numainfo;
1103          struct xen_sysctl_sched_id          sched_id;
1104          struct xen_sysctl_perfc_op          perfc_op;
1105          struct xen_sysctl_getdomaininfolist getdomaininfolist;
1106          struct xen_sysctl_debug_keys        debug_keys;
1107          struct xen_sysctl_getcpuinfo        getcpuinfo;
1108          struct xen_sysctl_availheap         availheap;
1109          struct xen_sysctl_get_pmstat        get_pmstat;
1110          struct xen_sysctl_cpu_hotplug       cpu_hotplug;
1111          struct xen_sysctl_pm_op             pm_op;
1112          struct xen_sysctl_page_offline_op   page_offline;
1113          struct xen_sysctl_lockprof_op       lockprof_op;
1114          struct xen_sysctl_cpupool_op        cpupool_op;
1115          struct xen_sysctl_scheduler_op      scheduler_op;
1116          struct xen_sysctl_coverage_op       coverage_op;
1117          struct xen_sysctl_psr_cmt_op        psr_cmt_op;
1118          struct xen_sysctl_psr_alloc         psr_alloc;
1119          struct xen_sysctl_cpu_levelling_caps cpu_levelling_caps;
1120          struct xen_sysctl_cpu_featureset    cpu_featureset;
1121          struct xen_sysctl_livepatch_op      livepatch;
1122  #if defined(__i386__) || defined(__x86_64__)
1123          struct xen_sysctl_cpu_policy        cpu_policy;
1124  #endif
1125          uint8_t                             pad[128];
1126      } u;
1127  };
1128  typedef struct xen_sysctl xen_sysctl_t;
1129  DEFINE_XEN_GUEST_HANDLE(xen_sysctl_t);
1130  
1131  #endif /* __XEN_PUBLIC_SYSCTL_H__ */
1132  
1133  /*
1134   * Local variables:
1135   * mode: C
1136   * c-file-style: "BSD"
1137   * c-basic-offset: 4
1138   * tab-width: 4
1139   * indent-tabs-mode: nil
1140   * End:
1141   */
1142