zx_object_get_info

摘要

查询对象的相关信息。

声明

#include <zircon/syscalls.h>

zx_status_t zx_object_get_info(zx_handle_t handle,
                               uint32_t topic,
                               void* buffer,
                               size_t buffer_size,
                               size_t* actual,
                               size_t* avail);

说明

zx_object_get_info() 会请求提供的句柄(或句柄所引用的对象)的相关信息。topic 参数用于指明所需的具体信息。

buffer 是指向大小为 buffer_size 的缓冲区的指针,用于返回信息。

actual 是一个可选指针,用于返回已写入缓冲区的记录数。

avail 是一个可选指针,用于返回可读取的记录数。

如果缓冲区大小不足,avail 将大于 actual

主题

ZX_INFO_HANDLE_VALID

handle 类型:Any

缓冲区类型:n/a

如果 handle 有效,则返回 ZX_OK;否则,返回 ZX_ERR_BAD_HANDLE。系统不会返回任何记录,并且 buffer 可能为 NULL。

ZX_INFO_HANDLE_BASIC

handle 类型:Any

缓冲区类型:zx_info_handle_basic_t[1]

typedef struct zx_info_handle_basic {
    // The unique id assigned by kernel to the object referenced by the
    // handle.
    zx_koid_t koid;

    // The immutable rights assigned to the handle. Two handles that
    // have the same koid and the same rights are equivalent and
    // interchangeable.
    zx_rights_t rights;

    // The object type: channel, event, socket, etc.
    uint32_t type;                // zx_obj_type_t;

    // If the object referenced by the handle is related to another (such
    // as the other end of a channel, or the parent of a job) then
    // |related_koid| is the koid of that object, otherwise it is zero.
    // This relationship is immutable: an object's |related_koid| does
    // not change even if the related object no longer exists.
    zx_koid_t related_koid;
} zx_info_handle_basic_t;

ZX_INFO_HANDLE_COUNT

handle 类型:Any

缓冲区类型:zx_info_handle_count_t[1]

typedef struct zx_info_handle_count {
    // The number of outstanding handles to a kernel object.
    uint32_t handle_count;
} zx_info_handle_count_t;

handle_count 应仅用作调试辅助。请勿使用它来检查不受信任的进程能否修改内核对象。由于系统调度程序的异步性质,在最后一个句柄从一个进程传输到另一个进程期间,可能存在一个时间窗口,在此期间,上一个句柄所有者可能会修改对象。

ZX_INFO_PROCESS_HANDLE_STATS

handle 类型:Process

缓冲区类型:zx_info_process_handle_stats_t[1]

typedef struct zx_info_process_handle_stats {
    // The number of outstanding handles to kernel objects of each type.
    uint32_t handle_count[ZX_OBJ_TYPE_UPPER_BOUND];
} zx_info_process_handle_stats_t;

ZX_INFO_HANDLE_TABLE

handle 类型:Process

缓冲区类型:zx_info_handle_extended_t[n]

返回一个 zx_info_handle_extended_t 数组,其中包含调用时进程中的每个句柄。内核会确保返回的句柄保持一致。

typedef struct zx_info_handle_extended {
    // The object type: channel, event, socket, etc.
    zx_obj_type_t type;

    // The handle value, which is only valid for the process that
    // was passed to ZX_INFO_HANDLE_TABLE.
    zx_handle_t handle_value;

    // The immutable rights assigned to the handle. Two handles that
    // have the same koid and the same rights are equivalent and
    // interchangeable.
    zx_rights_t rights;

    uint32_t reserved;

    // The unique id assigned by kernel to the object referenced by the
    // handle.
    zx_koid_t koid;

    // If the object referenced by the handle is related to another (such
    // as the other end of a channel, or the parent of a job) then
    // |related_koid| is the koid of that object, otherwise it is zero.
    // This relationship is immutable: an object's |related_koid| does
    // not change even if the related object no longer exists.
    zx_koid_t related_koid;

    // If the object referenced by the handle has a peer, like the
    // other end of a channel, then this is the koid of the process
    // which currently owns it.
    zx_koid_t peer_owner_koid;
} zx_info_handle_extended_t;

请注意,进程可能对其没有句柄的对象具有有效引用。例如,所有句柄均已关闭的正在运行的线程。

ZX_INFO_JOB

handle 类型:Job

缓冲区类型:zx_info_job_t[1]

typedef struct zx_info_job {
    // The job's return code; only valid if |exited| is true.
    // If the code is valid, it will be one of the ZX_TASK_RETCODE values.
    int64_t return_code;

    // If true, the job has exited and |return_code| is valid.
    // Killing a job is the only way for a job to exit.
    bool exited;

    // True if the ZX_PROP_JOB_KILL_ON_OOM property was set.
    bool kill_on_oom;

    // True if a debugger is attached to the job.
    bool debugger_attached;
} zx_info_job_t;

请注意,|exited| 会在 |zx_task_kill| 或等效操作(例如 OOM 终止)后立即报告作业已退出,但子作业和进程可能仍在退出过程中。

ZX_INFO_PROCESS

handle 类型:Process

缓冲区类型:zx_info_process_t[1]

typedef struct zx_info_process {
    // The process's return code; only valid if the
    // |ZX_PROCESS_INFO_FLAG_EXITED| flag is set. If the process was killed, it
    // will be one of the |ZX_TASK_RETCODE| values.
    int64_t return_code;

    // The monotonic time at which `zx_process_start()` was called, only valid
    // if the |ZX_INFO_PROCESS_FLAG_STARTED| flag is set.
    zx_instant_mono_t start_time;

    // Bitwise OR of ZX_INFO_PROCESS_FLAG_* values.
    uint32_t flags;
} zx_info_process_t;

请注意,|flags| 会立即报告进程在执行 |zx_task_kill| 后已退出(即,其中将包含 ZX_INFO_PROCESS_FLAG_EXITED),但子线程可能仍在退出进程中。

ZX_INFO_PROCESS_THREADS

handle 类型:Process

buffer 类型:zx_koid_t[n]

返回一个 zx_koid_t 数组,其中每个元素对应于进程中此时正在运行的每个线程。

请注意,获取线程列表本身就存在争用问题。可以先暂停所有线程来稍微缓解此问题,但请注意,外部线程可以创建新线程。actual 将包含 buffer 中返回的线程数。avail 将包含获取线程列表时进程的线程总数,该值可能大于 actual

ZX_INFO_THREAD

handle 类型:Thread

缓冲区类型:zx_info_thread_t[1]

typedef struct zx_info_thread {
    // One of ZX_THREAD_STATE_* values.
    uint32_t state;

    // If |state| is ZX_THREAD_STATE_BLOCKED_EXCEPTION, the thread has gotten
    // an exception and is waiting for the exception to be handled by the
    // specified channel.
    // The value is one of ZX_EXCEPTION_CHANNEL_TYPE_*.
    uint32_t wait_exception_channel_type;

    // CPUs this thread may be scheduled on, as specified by
    // a profile object applied to this thread.
    //
    // The kernel may not internally store invalid CPUs in the mask, so
    // this may not exactly match the mask applied to the thread for
    // CPUs beyond what the system is able to use.
    zx_cpu_set_t cpu_affinity_mask;
} zx_info_thread_t;

目前,此结构体中的值主要用于信息和调试目的。

各种 ZX_THREAD_STATE_ 值由以下代码定义

#include <zircon/syscalls/object.h>
  • ZX_THREAD_STATE_NEW:线程已创建,但尚未开始运行。
  • ZX_THREAD_STATE_RUNNING:线程正在正常运行用户代码。
  • ZX_THREAD_STATE_SUSPENDED:因 zx_task_suspend() 而停止。
  • ZX_THREAD_STATE_BLOCKED:在系统调用中或处理异常时。此值绝不会单独返回。请参阅下文中的 `ZX_THREAD_STATEBLOCKED`*。
  • ZX_THREAD_STATE_DYING:线程正在终止过程中,但尚未停止。
  • ZX_THREAD_STATE_DEAD:线程已停止运行。

当线程在阻塞调用内停止或在异常中停止时,state 中返回的值为以下之一:

各种 ZX_EXCEPTION_CHANNEL_TYPE_ 值由以下代码定义

#include <zircon/syscalls/exception.h>
  • ZX_EXCEPTION_CHANNEL_TYPE_NONE
  • ZX_EXCEPTION_CHANNEL_TYPE_DEBUGGER
  • ZX_EXCEPTION_CHANNEL_TYPE_THREAD
  • ZX_EXCEPTION_CHANNEL_TYPE_PROCESS
  • ZX_EXCEPTION_CHANNEL_TYPE_JOB
  • ZX_EXCEPTION_CHANNEL_TYPE_JOB_DEBUGGER

ZX_INFO_THREAD_EXCEPTION_REPORT

handle 类型:Thread

缓冲区类型:zx_exception_report_t[1]

#include <zircon/syscalls/exception.h>

如果线程当前处于异常中并正在等待异常响应,则会以单个 zx_exception_report_t 的形式返回异常报告,其状态为 ZX_OK

如果线程未处于异常状态且正在等待异常响应,则返回 ZX_ERR_BAD_STATE

ZX_INFO_THREAD_STATS

handle 类型:Thread

buffer 类型:zx_info_thread_stats[1]

typedef struct zx_info_thread_stats {
    // Total accumulated running time of the thread.
    //
    // Note: See zx_info_task_runtime for queue time in addition to runtime.
    zx_duration_t total_runtime;

    // CPU number that this thread was last scheduled on, or ZX_INFO_INVALID_CPU
    // if the thread has never been scheduled on a CPU. By the time this call
    // returns, the thread may have been scheduled elsewhere, so this
    // information should only be used as a hint or for statistics.
    uint32_t last_scheduled_cpu;
} zx_info_thread_stats_t;

如果线程已退出,则返回 ZX_ERR_BAD_STATE

ZX_INFO_GUEST_STATS

handle 类型:Resource(具体来说是 info 资源)

缓冲区类型:zx_info_guest_stats_t[1]

// Each machine has its own format for the same ZX_INFO_GUEST_STATS topic.
// In each build, zx_info_guest_stats_t is a typedef alias for the type.
// Cross-tools can select the machine-specific type to use based on the
// source of the data they are working with.
typedef struct zx_arm64_info_guest_stats {
    uint32_t cpu_number;
    uint32_t flags;
    uint64_t vm_entries;
    uint64_t vm_exits;
    uint64_t wfi_wfe_instructions;
    uint64_t instruction_aborts;
    uint64_t data_aborts;
    uint64_t system_instructions;
    uint64_t smc_instructions;
    uint64_t interrupts;
} zx_arm64_info_guest_stats_t;

typedef struct zx_x86_64_info_guest_stats {
    uint32_t cpu_number;
    uint32_t flags;
    uint64_t vm_entries;
    uint64_t vm_exits;
    uint64_t interrupts;
    uint64_t interrupt_windows;
    uint64_t cpuid_instructions;
    uint64_t hlt_instructions;
    uint64_t control_register_accesses;
    uint64_t io_instructions;
    uint64_t rdmsr_instructions;
    uint64_t wrmsr_instructions;
    uint64_t ept_violations;
    uint64_t xsetbv_instructions;
    uint64_t pause_instructions;
    uint64_t vmcall_instructions;
} zx_x86_64_info_guest_stats;

ZX_INFO_CPU_STATS

handle 类型:Resource(具体而言,是 info 资源)

缓冲区类型:zx_info_cpu_stats_t[1]

typedef struct zx_info_cpu_stats {
    uint32_t cpu_number;
    uint32_t flags;

    zx_duration_t idle_time;

    // kernel scheduler counters
    uint64_t reschedules;
    uint64_t context_switches;
    uint64_t irq_preempts;
    uint64_t preempts;
    uint64_t yields;

    // cpu level interrupts and exceptions
    uint64_t ints;          // hardware interrupts, minus timer interrupts
                            // inter-processor interrupts
    uint64_t timer_ints;    // timer interrupts
    uint64_t timers;        // timer callbacks
    uint64_t page_faults;   // (deprecated, returns 0)
    uint64_t exceptions;    // (deprecated, returns 0)
    uint64_t syscalls;

    // inter-processor interrupts
    uint64_t reschedule_ipis;
    uint64_t generic_ipis;
} zx_info_cpu_stats_t;

ZX_INFO_VMAR

handle 类型:VM Address Region

缓冲区类型:zx_info_vmar_t[1]

typedef struct zx_info_vmar {
    // Base address of the region.
    uintptr_t base;

    // Length of the region, in bytes.
    size_t len;
} zx_info_vmar_t;

这会返回一个 zx_info_vmar_t,用于描述 VMAR 占用的地址空间范围。

ZX_INFO_VMAR_MAPS

handle 类型:Vm Address Region,使用 ZX_RIGHT_INSPECT

缓冲区类型:zx_info_maps_t[n]

zx_info_maps_t 数组是目标 VMAR 树的深度优先预订遍历。根据预序遍历,基地址将按升序排列。

如需了解 zx_info_maps_t,请参阅 ZX_INFO_PROCESS_MAPS

第一个 zx_info_maps_t 将描述查询的 VMAR。每个条目的 depth 字段描述其与前面的节点的关系。所查询的 VMAR 的深度为 0。所有其他条目的深度均为 1 或更大。

其他错误:

  • ZX_ERR_ACCESS_DENIED:缺少适当权限。
  • ZX_ERR_BAD_STATE:如果 VMAR 或包含 VMAR 的地址空间已被销毁,或者包含 VMAR 的进程已终止。

ZX_INFO_VMO

handle 类型:VM Object

缓冲区类型:zx_info_vmo_t[1]

typedef struct zx_info_vmo {
    // The koid of this VMO.
    zx_koid_t koid;

    // The name of this VMO.
    char name[ZX_MAX_NAME_LEN];

    // The size of this VMO; i.e., the amount of virtual address space it
    // would consume if mapped.
    uint64_t size_bytes;

    // If this VMO is a child , the koid of its parent. Otherwise, zero.
    // See |flags| for the type of child.
    zx_koid_t parent_koid;

    // The number of children of this VMO, if any.
    size_t num_children;

    // The number of times this VMO is currently mapped into VMARs.
    // Note that the same process will often map the same VMO twice,
    // and both mappings will be counted here. (I.e., this is not a count
    // of the number of processes that map this VMO; see share_count.)
    size_t num_mappings;

    // An estimate of the number of unique address spaces that
    // this VMO is mapped into. Every process has its own address space,
    // and so does the kernel.
    size_t share_count;

    // Bitwise OR of ZX_INFO_VMO_* values.
    uint32_t flags;

    // If |ZX_INFO_VMO_TYPE(flags) == ZX_INFO_VMO_TYPE_PAGED|, the amount of
    // memory currently allocated to this VMO; i.e., the amount of physical
    // memory it consumes. Undefined otherwise.
    uint64_t committed_bytes;

    // If |flags & ZX_INFO_VMO_VIA_HANDLE|, the handle rights.
    //
    // If |flags & ZX_INFO_VMO_VIA_IOB_HANDLE|, the effective combined
    // handle rights for the IOB region and containing IOB.
    //
    // Undefined otherwise.
    zx_rights_t handle_rights;

    // VMO mapping cache policy. One of ZX_CACHE_POLICY_*
    uint32_t cache_policy;

    // Amount of kernel memory, in bytes, allocated to track metadata
    // associated with this VMO.
    uint64_t metadata_bytes;

    // Running counter of the number of times the kernel, without user request,
    // performed actions on this VMO that would have caused |committed_bytes| to
    // report a different value.
    uint64_t committed_change_events;

    // If |ZX_INFO_VMO_TYPE(flags) == ZX_INFO_VMO_TYPE_PAGED|, the amount of
    // content that has been populated and is being tracked by this vmo. This
    // can be greater than |committed_bytes| where content might be compressed
    // or otherwise tracked in a way that does not correlate directly to being
    // committed.
    uint64_t populated_bytes;
} zx_info_vmo_t;

这会返回一个 zx_info_vmo_t,用于描述 VMO 的各种属性。

ZX_INFO_SOCKET

handle 类型:Socket

缓冲区类型:zx_info_socket_t[1]

typedef struct zx_info_socket {
    // The options passed to zx_socket_create().
    uint32_t options;

    // The maximum size of the receive buffer of a socket, in bytes.
    //
    // The receive buffer may become full at a capacity less than the maximum
    // due to overhead.
    size_t rx_buf_max;

    // The size of the receive buffer of a socket, in bytes.
    size_t rx_buf_size;

    // The amount of data, in bytes, that is available for reading in a single
    // zx_socket_read call.
    //
    // For stream sockets, this value will match |rx_buf_size|. For datagram
    // sockets, this value will be the size of the next datagram in the receive
    // buffer.
    size_t rx_buf_available;

    // The maximum size of the transmit buffer of a socket, in bytes.
    //
    // The transmit buffer may become full at a capacity less than the maximum
    // due to overhead.
    //
    // Will be zero if the peer endpoint is closed.
    size_t tx_buf_max;

    // The size of the transmit buffer of a socket, in bytes.
    //
    // Will be zero if the peer endpoint is closed.
    size_t tx_buf_size;
} zx_info_socket_t;

ZX_INFO_TIMER

handle 类型:Timer

缓冲区类型:zx_info_timer_t[1]

typedef struct zx_info_timer {
    // The options passed to zx_timer_create().
    uint32_t options;

    // The deadline with respect to ZX_CLOCK_MONOTONIC at which the timer will
    // fire next.
    //
    // This value will be zero if the timer is not set to fire.
    zx_time_t deadline;

    // Specifies a range from deadline - slack to deadline + slack during which
    // the timer is allowed to fire. The system uses this parameter as a hint to
    // coalesce nearby timers.
    //
    // The precise coalescing behavior is controlled by the options parameter
    // specified when the timer was created.
    //
    // This value will be zero if the timer is not set to fire.
    zx_duration_t slack;
} zx_info_timer_t;

ZX_INFO_JOB_CHILDREN

handle 类型:Job

缓冲区类型:zx_koid_t[n]

返回一个 zx_koid_t 数组,其中每个 zx_koid_t 对应于所提供作业句柄的每个直接子作业。

ZX_INFO_JOB_PROCESSES

handle 类型:Job

缓冲区类型:zx_koid_t[n]

返回一个 zx_koid_t 数组,其中每个 zx_koid_t 对应于所提供作业句柄的每个直接子进程。

ZX_INFO_TASK_STATS

handle 类型:Process

buffer 类型:zx_info_task_stats_t[1]

返回有关任务使用的资源(例如内存)的统计信息。

typedef struct zx_info_task_stats {
    // The total size of mapped memory ranges in the task.
    // Not all will be backed by physical memory.
    size_t mem_mapped_bytes;

    // For the fields below, a byte is considered committed if it's backed by
    // physical memory. Some of the memory may be double-mapped, and thus
    // double-counted.

    // Committed memory that is only mapped into this task.
    size_t mem_private_bytes;

    // Committed memory that is mapped into this and at least one other task.
    size_t mem_shared_bytes;

    // A number that estimates the fraction of mem_shared_bytes that this
    // task is responsible for keeping alive.
    //
    // An estimate of:
    //   For each shared, committed byte:
    //   mem_scaled_shared_bytes += 1 / (number of tasks mapping this byte)
    //
    // This number is strictly smaller than mem_shared_bytes.
    size_t mem_scaled_shared_bytes;
} zx_info_task_stats_t;

其他错误:

  • ZX_ERR_BAD_STATE:如果目标进程已终止

ZX_INFO_TASK_RUNTIME

handle 类型:JobProcessThread

缓冲区类型:zx_info_task_runtime_t[1]

返回有关任务运行时的统计信息。

// Info on the runtime of a task.
typedef struct zx_info_task_runtime {
    // The total amount of time this task and its children were
    // running on a CPU (not blocked).
    // * Threads include only their own runtime.
    // * Processes include the runtime for all of their threads (including threads that previously
    // exited).
    // * Jobs include the runtime for all of their processes (including processes that previously
    // exited).
    zx_duration_t cpu_time;

    // The total amount of time this task and its children were queued
    // to run (ready) but not actually using a CPU.
    // * Threads include only their own queue time.
    // * Processes include the queue time for all of their threads (including threads that
    // previously exited).
    // * Jobs include the queue time for all of their processes (including processes that previously
    // exited).
    zx_duration_t queue_time;

    // The total amount of time this task and its children spent handling page faults.
    // * Threads include only their own page fault handling time.
    // * Processes include the page fault time for all of their threads (including threads that
    // previously exited).
    // * Jobs include the page fault time for all of their processes (including processes that
    // previously exited).
    zx_duration_t page_fault_time;

    // The total amount of time this task and its children spent waiting on contended kernel locks.
    // * Threads include only their own wait time.
    // * Processes include the wait time for all of their threads (including threads that
    // previously exited).
    // * Jobs include the wait time for all of their processes (including processes that
    // previously exited).
    zx_duration_t lock_contention_time;
} zx_info_task_runtime_t;

任务的运行时间不包括在挂起或阻塞状态下等待事件或 I/O 所花的时间。这些统计信息可能会用于:

  1. 估算任务已使用的 CPU 时间。
  2. 估算任务因其他任务(队列时间)、页面故障处理程序和内核锁争用而导致的延迟时间。

ZX_INFO_PROCESS_MAPS

handle 类型:Process,包含 ZX_RIGHT_READ

buffer 类型:zx_info_maps_t[n]

zx_info_maps_t 数组是目标进程的 Aspace/VMAR/Mapping 树的深度优先预顺序遍历。根据预序遍历,基地址将按升序排列。

typedef struct zx_info_maps {
    // Name if available; empty string otherwise.
    char name[ZX_MAX_NAME_LEN];
    // Base address.
    zx_vaddr_t base;
    // Size in bytes.
    size_t size;

    // The depth of this node in the tree.
    // Can be used for indentation, or to rebuild the tree from an array
    // of zx_info_maps_t entries, which will be in depth-first pre-order.
    size_t depth;
    // The type of this entry; indicates which union entry is valid.
    uint32_t type; // zx_info_maps_type_t
    union {
        zx_info_maps_mapping_t mapping;
        // No additional fields for other types.
    } u;
} zx_info_maps_t;

typedef struct zx_info_maps_mapping {
    // MMU flags for the mapping.
    // Bitwise OR of ZX_VM_PERM_{READ,WRITE,EXECUTE} values.
    zx_vm_option_t mmu_flags;
    uint8_t padding1[4];
    // koid of the mapped VMO or IOB region.
    zx_koid_t vmo_koid;
    // Offset into the above VMO or IOB region.
    uint64_t vmo_offset;
    // The number of PAGE_SIZE pages in the mapped region of the VMO or
    // IOB region that are backed by physical memory.
    size_t committed_pages;
    // The number of PAGE_SIZE pages of content that have been populated and are
    // being tracked in the mapped region of the VMO or IOB region. This can be
    // greater than |committed_pages| where pages might be compressed or otherwise
    // tracked in a way that does not correlate directly to being committed.
    size_t populated_pages;
} zx_info_maps_mapping_t;

每个条目的 depth 字段说明了其与前面的节点的关系。深度 0 是根 Aspace,深度 1 是根 VMAR,所有其他条目的深度均为 2 或更大。

要全面了解进程如何使用其 VMO 以及各种进程如何使用 VMO,您可能需要将这些信息与 ZX_INFO_PROCESS_VMOS 结合使用。

如需查看此主题的用户示例,并按 koid 转储任意进程的映射,请参阅 vmaps 命令行工具。

其他错误:

  • ZX_ERR_ACCESS_DENIED:缺少适当权限。
  • ZX_ERR_BAD_STATE:如果目标进程已终止,或者其地址空间已被销毁

ZX_INFO_PROCESS_VMOS

handle 类型:Process,使用 ZX_RIGHT_READ

缓冲区类型:zx_info_vmo_t[n]

zx_info_vmo_t 数组是目标进程指向的所有 VMO 的列表。有些 VMO 经过映射,有些由 VMO 或 IOB 句柄指向,还有一些则是这些记录的组合。返回的结构体的 flags 字段将指示 ZX_INFO_VMO_VIA_HANDLE、ZX_INFO_VMO_VIA_IOB_HANDLE 或 ZX_INFO_VMO_VIA_MAPPING 之一,以便进行区分。

如需全面了解进程如何使用其 VMO 以及各种进程如何使用 VMO,您可能需要将此信息与 ZX_INFO_PROCESS_MAPS 结合使用。

// Describes a VMO.
typedef struct zx_info_vmo {
    // The koid of this VMO.
    zx_koid_t koid;

    // The name of this VMO.
    char name[ZX_MAX_NAME_LEN];

    // The size of this VMO; i.e., the amount of virtual address space it
    // would consume if mapped.
    uint64_t size_bytes;

    // If this VMO is a child , the koid of its parent. Otherwise, zero.
    // See |flags| for the type of child.
    zx_koid_t parent_koid;

    // The number of children of this VMO, if any.
    size_t num_children;

    // The number of times this VMO is currently mapped into VMARs.
    // Note that the same process will often map the same VMO twice,
    // and both mappings will be counted here. (I.e., this is not a count
    // of the number of processes that map this VMO; see share_count.)
    size_t num_mappings;

    // An estimate of the number of unique address spaces that
    // this VMO is mapped into. Every process has its own address space,
    // and so does the kernel.
    size_t share_count;

    // Bitwise OR of ZX_INFO_VMO_* values.
    uint32_t flags;

    // If |ZX_INFO_VMO_TYPE(flags) == ZX_INFO_VMO_TYPE_PAGED|, the amount of
    // memory currently allocated to this VMO; i.e., the amount of physical
    // memory it consumes. Undefined otherwise.
    uint64_t committed_bytes;

    // If |flags & ZX_INFO_VMO_VIA_HANDLE|, the handle rights.
    //
    // If |flags & ZX_INFO_VMO_VIA_IOB_HANDLE|, the effective combined
    // handle rights for the IOB region and containing IOB.
    //
    // Undefined otherwise.
    zx_rights_t handle_rights;

    // VMO mapping cache policy. One of ZX_CACHE_POLICY_*
    uint32_t cache_policy;

    // Amount of kernel memory, in bytes, allocated to track metadata
    // associated with this VMO.
    uint64_t metadata_bytes;

    // Running counter of the number of times the kernel, without user request,
    // performed actions on this VMO that would have caused |committed_bytes| to
    // report a different value.
    uint64_t committed_change_events;

    // If |ZX_INFO_VMO_TYPE(flags) == ZX_INFO_VMO_TYPE_PAGED|, the amount of
    // content that has been populated and is being tracked by this vmo. This
    // can be greater than |committed_bytes| where content might be compressed
    // or otherwise tracked in a way that does not correlate directly to being
    // committed.
    uint64_t populated_bytes;
} zx_info_vmo_t;

如需查看此主题的用户示例,并按 koid 转储任意进程的 VMO,请参阅 vmos 命令行工具。

ZX_INFO_KMEM_STATS

handle 类型:Resource(具体而言,是 info 资源)

缓冲区类型:zx_info_kmem_stats_t[1]

返回内核看到的内存用量信息。

typedef struct zx_info_kmem_stats {
    // The total amount of physical memory available to the system.
    // Note, the values below may not exactly add up to this total.
    uint64_t total_bytes;

    // The amount of unallocated memory available for general use. This is a
    // subset of |total_bytes|.
    uint64_t free_bytes;

    // The amount of unallocated memory loaned from VMOs that is available for
    // allocations that support loaned memory. This is a subset of
    // |total_bytes| and does not overlap with |free_bytes|.
    uint64_t free_loaned_bytes;

    // The amount of memory reserved by and mapped into the kernel for reasons
    // not covered by other fields in this struct. Typically for readonly data
    // like the ram disk and kernel image, and for early-boot dynamic memory.
    // This value of this field should not typically change post boot and is a
    // subset of |total_bytes|.
    uint64_t wired_bytes;

    // The amount of memory allocated to the general kernel heap. This is a
    // subset of |total_bytes|.
    uint64_t total_heap_bytes;

    // The portion of |total_heap_bytes| that is not holding an allocated
    // object.
    uint64_t free_heap_bytes;

    // The amount of memory committed to VMOs created by both kernel and user.
    // Does not include certain VMOs that fall under |wired_bytes|. This is a
    // subset of |total_bytes|.
    uint64_t vmo_bytes;

    // The amount of memory used for architecture-specific MMU metadata
    // like page tables for both kernel and user mappings. This is a subset of
    // |total_bytes|.
    uint64_t mmu_overhead_bytes;

    // The amount of memory in use by IPC. This is a subset of |total_bytes|.
    uint64_t ipc_bytes;

    // The amount of memory in use by kernel allocation caches. This memory is
    // not allocated, but is only available for use for specific kernel
    // allocation requests. This is a subset of |total_bytes|.
    uint64_t cache_bytes;

    // The amount of memory in use by the kernel in slab allocators for kernel
    // objects. Unlike the heap there is no measurement for the amount of slab
    // memory that is not presently in use. This is a subset of |total_bytes|.
    uint64_t slab_bytes;

    // The amount of memory in use for storing compressed data that would
    // otherwise be part of VMOs.
    // Use ZX_INFO_KMEM_STATS_COMPRESSION for more details. This is a subset of
    // |total_bytes|.
    uint64_t zram_bytes;

    // Non-free memory that isn't accounted for in any other field. This is a
    // subset of |total_bytes|.
    uint64_t other_bytes;

    // The amount of memory committed to VMOs that is reclaimable by the kernel.
    // This is a subset of |vmo_bytes|.
    uint64_t vmo_reclaim_total_bytes;

    // The amount of memory committed to reclaimable VMOs, that has been most
    // recently accessed, and would not be eligible for eviction by the kernel
    // under memory pressure. This is a subset of |vmo_reclaim_total_bytes|.
    uint64_t vmo_reclaim_newest_bytes;

    // The amount of memory committed to reclaimable VMOs, that has been least
    // recently accessed, and would be the first to be evicted by the kernel
    // under memory pressure. This is a subset of |reclaim_total_bytes|.
    uint64_t vmo_reclaim_oldest_bytes;

    // The amount of memory in VMOs that would otherwise be tracked for
    // reclamation, but has had reclamation disabled. This is a subset of
    // |vmo_bytes|.
    uint64_t vmo_reclaim_disabled_bytes;

    // The amount of memory committed to discardable VMOs that is currently
    // locked, or unreclaimable by the kernel under memory pressure. This is a
    // subset of |vmo_bytes| and some of this count may be included in any other
    // |vmo_reclaim_*| count.
    uint64_t vmo_discardable_locked_bytes;

    // The amount of memory committed to discardable VMOs that is currently
    // unlocked, or reclaimable by the kernel under memory pressure. This is a
    // subset of |vmo_bytes| and some of this count may be included in any other
    // |vmo_reclaim_*| count
    uint64_t vmo_discardable_unlocked_bytes;
 } zx_info_kmem_stats_t;

ZX_INFO_KMEM_STATS_COMPRESSION

handle 类型:Resource(具体而言,是 info 资源)

buffer 类型:zx_info_kmem_stats_compression_t[1]

返回与内核压缩内存子系统相关的内存用量信息。

typedef struct zx_info_kmem_stats_compression {
    // Size in bytes of the content that is currently being compressed and stored.
    uint64_t uncompressed_storage_bytes;

    // Size in bytes of all memory, including metadata, fragmentation and other
    // overheads, of the compressed memory area. Note that due to base book
    // keeping overhead this could be non-zero, even when
    // |uncompressed_content_bytes| is zero.
    uint64_t compressed_storage_bytes;

    // Size in bytes of any fragmentation in the compressed memory area.
    uint64_t compressed_fragmentation_bytes;

    // Total amount of CPU time spent on compression across all threads.
    // Compression may happen in parallel and so this can be larger than
    // wall clock time.
    zx_duration_t compression_time;

    // Total amount of time decompression has spent on a CPU across all threads.
    // Decompression may happen in parallel and so this can increase faster than
    // wall clock time.
    zx_duration_t decompression_time;

    // Total number of times compression has been done on a page, regardless of
    // whether the compressed result was ultimately retained.
    uint64_t total_page_compression_attempts;

    // How many of the total compression attempts were considered failed and
    // were not stored. An example reason for failure would be a page not being
    // compressed sufficiently to be considered worth storing.
    uint64_t failed_page_compression_attempts;

    // Number of times pages have been decompressed.
    uint64_t total_page_decompressions;

    // Number of times a page was removed from storage without needing to be
    // decompressed. An example that would cause this is a VMO being destroyed.
    uint64_t compressed_page_evictions;

    // How many pages compressed due to the page being inactive, but without
    // there being memory pressure.
    uint64_t eager_page_compressions;

    // How many pages compressed due to general memory pressure. This excludes pages
    // compressed due to critical memory pressure.
    uint64_t memory_pressure_page_compressions;

    // How many pages compressed due to attempting to avoid OOM or near OOM
    // scenarios.
    uint64_t critical_memory_page_compressions;

    // The nanoseconds in the base unit of time for
    // |pages_decompressed_within_log_time|.
    uint64_t pages_decompressed_unit_ns;

    // How long pages spent compressed before being decompressed, grouped in log
    // buckets. Pages that got evicted, and hence were not decompressed, are not
    // counted here. Buckets are in |pages_decompressed_unit_ns| and round up
    // such that:
    // 0: Pages decompressed in <1 unit
    // 1: Pages decompressed between 1 and 2 units
    // 2: Pages decompressed between 2 and 4 units
    // ...
    // 7: Pages decompressed between 64 and 128 units
    // How many pages are held compressed for longer than 128 units can be
    // inferred by subtracting from |total_page_decompressions|.
    uint64_t pages_decompressed_within_log_time[8];
} zx_info_kmem_stats_compression_t;

ZX_INFO_RESOURCE

handle 类型:Resource

buffer 类型:zx_info_resource_t[1]

通过资源对象的句柄返回资源对象的相关信息。

typedef struct zx_info_resource {
    // The resource kind; resource object kinds are described in resource.md
    uint32_t kind;
    // Resource's creation flags
    uint32_t flags;
    // Resource's base value (inclusive)
    uint64_t base;
    // Resource's length value
    size_t size;
    char name[ZX_MAX_NAME_LEN];
} zx_info_resource_t;

资源种类是以下各项之一

  • ZX_RSRC_KIND_ROOT
  • ZX_RSRC_KIND_MMIO
  • ZX_RSRC_KIND_IOPORT
  • ZX_RSRC_KIND_IRQ
  • ZX_RSRC_KIND_SMC
  • ZX_RSRC_KIND_SYSTEM

ZX_INFO_BTI

handle 类型:Bus Transaction Initiator

buffer 类型:zx_info_bti_t[1]

typedef struct zx_info_bti {
    // zx_bti_pin will always be able to return addresses that are contiguous for at
    // least this many bytes. E.g. if this returns 1MB, then a call to
    // zx_bti_pin() with a size of 2MB will return at most two physically-contiguous runs.
    // If the size were 2.5MB, it will return at most three physically-contiguous runs.
    uint64_t minimum_contiguity;

    // The number of bytes in the device's address space (UINT64_MAX if 2^64).
    uint64_t aspace_size;

    // The count of the pinned memory object tokens. Requesting this count is
    // racy, so this should only be used for informative reasons.
    uint64_t pmo_count;

    // The count of the quarantined pinned memory object tokens. Requesting this count is
    // racy, so this should only be used for informative reasons.
    uint64_t quarantine_count;
} zx_info_bti_t;

主题 ZX_INFO_IOB

handle 类型:IOBuffer

缓冲区类型:zx_info_iob_t[1]

返回有关整个 IOB 实例的信息。

typedef struct zx_info_iob {
  // The value of the *options* parameter passed to `zx_iob_create`.
  uint64_t options;
  // The number of regions in the IOB.
  uint32_t region_count;
  // Reserved for future extensions.
  uint8_t padding[4];
} zx_info_iob_t;

主题 ZX_INFO_IOB_REGIONS

zx_iob_region_info_t 数组的形式返回 IOB 的每个区域的信息

handle 类型:IOBuffer

缓冲区类型:zx_iob_region_info_t[n]

struct zx_iob_region_info_t {
    /// The region description, with potentially swapped access bits.
    zx_iob_region_t region;
    /// The koid of the underlying memory object.
    zx_koid_t koid;
};

访问修饰符位会进行交换,以便 Ep0 访问位反映发出查询的端点的访问权限,而 Ep1 位反映另一个端点的访问权限,这样无需知道哪些句柄在创建时是 Ep0 和 Ep1,便可确定本地和远程句柄的访问权限。

ZX_INFO_POWER_DOMAINS

zx_power_domain_info_t 数组的形式返回每个已注册电源域的信息

handle 类型:Resource(具体而言,是 info 资源)

缓冲区类型:zx_power_domain_info_t[n]

 typedef struct zx_power_domain_info {
    /// CPUs part of this power domain.
    zx_cpu_set_t cpus;
    /// Id of the power domain.
    uint32_t domain_id;
    /// Number of idle power levels in this power domain.
    uint8_t idle_power_levels;
    /// Number of active power levels in this power domain.
    uint8_t active_power_levels;
    uint8_t padding1[2];
} zx_power_domain_info_t;

权限

如果主题ZX_INFO_PROCESS,则句柄必须为 ZX_OBJ_TYPE_PROCESS 类型且具有 ZX_RIGHT_INSPECT

如果 topicZX_INFO_JOB,则 handle 的类型必须为 ZX_OBJ_TYPE_JOB 且具有 ZX_RIGHT_INSPECT

如果 topicZX_INFO_PROCESS_THREADS,则 handle 的类型必须为 ZX_OBJ_TYPE_PROCESS 且具有 ZX_RIGHT_ENUMERATE

如果主题ZX_INFO_JOB_CHILDREN,则句柄必须为 ZX_OBJ_TYPE_JOB 类型且具有 ZX_RIGHT_ENUMERATE

如果主题ZX_INFO_JOB_PROCESSES,则句柄必须为 ZX_OBJ_TYPE_JOB 类型且具有 ZX_RIGHT_ENUMERATE

如果主题ZX_INFO_THREAD,则句柄必须为 ZX_OBJ_TYPE_THREAD 类型且具有 ZX_RIGHT_INSPECT

如果 topicZX_INFO_THREAD_EXCEPTION_REPORT,则 handle 的类型必须为 ZX_OBJ_TYPE_THREAD 且具有 ZX_RIGHT_INSPECT

如果 topicZX_INFO_THREAD_STATS,则 handle 的类型必须为 ZX_OBJ_TYPE_THREAD 且具有 ZX_RIGHT_INSPECT

如果 topicZX_INFO_TASK_STATS,则 handle 的类型必须为 ZX_OBJ_TYPE_PROCESS 且具有 ZX_RIGHT_INSPECT

如果 topicZX_INFO_PROCESS_MAPS,则 handle 的类型必须为 ZX_OBJ_TYPE_PROCESS 且具有 ZX_RIGHT_INSPECT

如果主题ZX_INFO_PROCESS_VMOS,则句柄必须为 ZX_OBJ_TYPE_PROCESS 类型且具有 ZX_RIGHT_INSPECT

如果 topicZX_INFO_VMO,则 handle 的类型必须为 ZX_OBJ_TYPE_VMO

如果主题ZX_INFO_VMAR,则句柄必须为 ZX_OBJ_TYPE_VMAR 类型且具有 ZX_RIGHT_INSPECT

如果主题ZX_INFO_VMAR_MAPS,则句柄必须为 ZX_OBJ_TYPE_VMAR 类型且具有 ZX_RIGHT_INSPECT

如果 topicZX_INFO_GUEST_STATShandle 必须具有基本 ZX_RSRC_SYSTEM_INFO_BASE 的资源种类 ZX_RSRC_KIND_SYSTEM

如果主题ZX_INFO_CPU_STATS句柄必须具有资源类型 ZX_RSRC_KIND_SYSTEM 和基础 ZX_RSRC_SYSTEM_INFO_BASE

如果主题ZX_INFO_KMEM_STATS句柄必须具有资源类型 ZX_RSRC_KIND_SYSTEM 和基础 ZX_RSRC_SYSTEM_INFO_BASE

如果主题ZX_INFO_KMEM_STATS_EXTENDED句柄必须具有资源类型 ZX_RSRC_KIND_SYSTEM 和基础 ZX_RSRC_SYSTEM_INFO_BASE

如果主题ZX_INFO_RESOURCE,则句柄必须为 ZX_OBJ_TYPE_RESOURCE 类型且具有 ZX_RIGHT_INSPECT

如果主题ZX_INFO_HANDLE_COUNT句柄必须包含 ZX_RIGHT_INSPECT

如果主题ZX_INFO_BTI,则句柄必须为 ZX_OBJ_TYPE_BTI 类型且具有 ZX_RIGHT_INSPECT

如果主题ZX_INFO_PROCESS_HANDLE_STATS,则句柄必须为 ZX_OBJ_TYPE_PROCESS 类型且具有 ZX_RIGHT_INSPECT

如果主题ZX_INFO_SOCKET,则句柄必须为 ZX_OBJ_TYPE_SOCKET 类型且具有 ZX_RIGHT_INSPECT

如果 topicZX_INFO_MSI,则 handle 的类型必须为 ZX_OBJ_TYPE_MSI 且具有 ZX_RIGHT_INSPECT

如果主题ZX_INFO_TASK_RUNTIME标识名的类型必须为 ZX_OBJ_TYPE_THREADZX_OBJ_TYPE_PROCESSZX_OBJ_TYPE_JOB,并且具有 ZX_RIGHT_INSPECT

如果主题ZX_INFO_POWER_DOMAINS句柄必须具有资源类型 ZX_RSRC_KIND_SYSTEM 和基础 ZX_RSRC_SYSTEM_INFO_BASE

返回值

zx_object_get_info() 会在成功时返回 ZX_OK。如果失败,则返回负错误值。

错误

ZX_ERR_BAD_HANDLE handle 不是有效的 handle。

ZX_ERR_WRONG_TYPE handle 不是topic 的适当类型

ZX_ERR_ACCESS_DENIED:如果 handle 不具备执行相应操作所需的权限。

ZX_ERR_INVALID_ARGS bufferactualavail 是无效的指针。

ZX_ERR_NO_MEMORY 因内存不足而失败。用户空间没有很好的方法来处理这种(不太可能发生的)错误。在未来的 build 中,此错误将不再出现。

ZX_ERR_BUFFER_TOO_SMALL 主题会返回固定数量的记录,但提供的缓冲区不足以存储这些记录。

ZX_ERR_NOT_SUPPORTED topic 不存在。

示例

bool is_handle_valid(zx_handle_t handle) {
    return zx_object_get_info(
        handle, ZX_INFO_HANDLE_VALID, NULL, 0, NULL, NULL) == ZX_OK;
}

zx_koid_t get_object_koid(zx_handle_t handle) {
    zx_info_handle_basic_t info;
    if (zx_object_get_info(handle, ZX_INFO_HANDLE_BASIC,
                           &info, sizeof(info), NULL, NULL) != ZX_OK) {
        return 0;
    }
    return info.koid;
}

void examine_threads(zx_handle_t proc) {
    zx_koid_t threads[128];
    size_t count, avail;

    if (zx_object_get_info(proc, ZX_INFO_PROCESS_THREADS, threads,
                           sizeof(threads), &count, &avail) != ZX_OK) {
        // Error!
    } else {
        if (avail > count) {
            // More threads than space in array;
            // could call again with larger array.
        }
        for (size_t n = 0; n < count; n++) {
            do_something(thread[n]);
        }
    }
}

另请参阅