/* Copyright (c) 2019-2022 Advanced Micro Devices, Inc.

 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal
 in the Software without restriction, including without limitation the rights
 to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 copies of the Software, and to permit persons to whom the Software is
 furnished to do so, subject to the following conditions:

 The above copyright notice and this permission notice shall be included in
 all copies or substantial portions of the Software.

 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 THE SOFTWARE. */

#ifndef AMD_DBGAPI_PROCESS_H
#define AMD_DBGAPI_PROCESS_H 1

#include "amd-dbgapi.h"
#include "callbacks.h"
#include "code_object.h"
#include "debug.h"
#include "dispatch.h"
#include "displaced_stepping.h"
#include "event.h"
#include "exception.h"
#include "handle_object.h"
#include "initialization.h"
#include "logging.h"
#include "os_driver.h"
#include "queue.h"
#include "utils.h"
#include "watchpoint.h"
#include "wave.h"

#include <cstddef>
#include <cstdint>
#include <functional>
#include <memory>
#include <optional>
#include <queue>
#include <string>
#include <tuple>
#include <type_traits>
#include <unordered_map>
#include <utility>
#include <variant>
#include <vector>

namespace amd::dbgapi
{

class process_t;

namespace detail
{

extern amd_dbgapi_callbacks_s process_callbacks;

/* The process that contained the last successful find result.  The next global
   find will start searching here.

   FIXME: If multi-process support is added this mechanism will need
   re-implementing to be thread safe.  */
extern process_t *last_found_process;

} /* namespace detail */

/* AMD Debugger API Process.  */

class process_t : public detail::handle_object<amd_dbgapi_process_id_t>
{
  static_assert (is_handle_type_v<amd_dbgapi_process_id_t>,
                 "amd_dbgapi_process_id_t is not a handle type");

public:
  enum class flag_t : uint32_t
  {
    /* The runtime was already enabled during attach.  There might already be
       active queues with waves executing.  */
    runtime_enable_during_attach = 1 << 0,
    /* Trap temporary register setup is now enabled.  Any new wave created
       after this flag is set will have its ttmps initialized by SPI.  */
    spi_ttmps_setup_enabled = 1 << 1
  };

private:
  static handle_object_set_t<process_t> s_process_map;

  amd_dbgapi_client_process_id_t const m_client_process_id;
  std::optional<amd_dbgapi_os_process_id_t> m_os_process_id{};

  os_runtime_info_t m_runtime_info{};
  amd_dbgapi_runtime_state_t m_runtime_state{
    AMD_DBGAPI_RUNTIME_STATE_UNLOADED
  };

  memory_cache_t m_memory_cache;
  std::unique_ptr<os_driver_t> m_os_driver{};
  flag_t m_flags{};

  os_wave_launch_mode_t m_wave_launch_mode{ os_wave_launch_mode_t::normal };
  os_wave_launch_trap_mask_t m_wave_trap_mask{};

  bool m_supports_precise_memory{ false };
  bool m_precise_memory{ false };

  bool m_forward_progress_needed{ true };

  pipe_t m_client_notifier_pipe{};

  std::unordered_map<os_watch_id_t, const watchpoint_t *> m_watchpoint_map{};

  std::queue<event_t *> m_pending_events{};

  std::tuple<
    handle_object_set_t<agent_t>, handle_object_set_t<breakpoint_t>,
    handle_object_set_t<code_object_t>, handle_object_set_t<dispatch_t>,
    handle_object_set_t<displaced_stepping_t>, handle_object_set_t<event_t>,
    handle_object_set_t<queue_t>, handle_object_set_t<watchpoint_t>,
    handle_object_set_t<wave_t>, handle_object_set_t<workgroup_t>>
    m_handle_object_sets{};

  const agent_t m_dummy_agent;

  std::pair<std::variant<process_t *, agent_t *, queue_t *>,
            os_exception_mask_t>
  query_debug_event (os_exception_mask_t exceptions_cleared);

public:
  process_t (amd_dbgapi_process_id_t process_id,
             amd_dbgapi_client_process_id_t client_process_id);
  ~process_t ();

  /* Disallow copying & moving process instances.  */
  process_t (const process_t &) = delete;
  process_t (process_t &&) = delete;
  process_t &operator= (const process_t &) = delete;
  process_t &operator= (process_t &&) = delete;

  /* Since process objects disallow copying & moving, two process objects are
     identical if they have the same address.  */
  bool operator== (const process_t &other) const { return this == &other; }
  bool operator!= (const process_t &other) const { return this != &other; }

  amd_dbgapi_client_process_id_t client_id () const
  {
    return m_client_process_id;
  }

  std::optional<amd_dbgapi_os_process_id_t> os_id () const
  {
    return m_os_process_id;
  }

  /* Reset all the handle_object_sets IDs.  There should not be any attached
     processes left in the s_process_map. */
  static void reset_all_ids ();

  memory_cache_t &memory_cache () { return m_memory_cache; }
  os_driver_t &os_driver () const { return *m_os_driver; }

  inline void set_flag (flag_t flags);
  inline void clear_flag (flag_t flags);
  inline bool is_flag_set (flag_t flags) const;

  [[nodiscard]] size_t
  read_global_memory_partial (amd_dbgapi_global_address_t address,
                              void *buffer, size_t size)
  {
    return m_memory_cache.read_global_memory (address, buffer, size);
  }

  [[nodiscard]] size_t
  write_global_memory_partial (amd_dbgapi_global_address_t address,
                               const void *buffer, size_t size)
  {
    return m_memory_cache.write_global_memory (address, buffer, size);
  }

  template <typename T>
  void read_global_memory (amd_dbgapi_global_address_t address, T *ptr,
                           size_t size = sizeof (T));
  template <typename T>
  void write_global_memory (amd_dbgapi_global_address_t address, const T *ptr,
                            size_t size = sizeof (T));

  void read_string (amd_dbgapi_global_address_t address, std::string *string,
                    size_t size);

  [[nodiscard]] size_t
  xfer_segment_memory (const address_space_t &address_space,
                       amd_dbgapi_segment_address_t segment_address,
                       void *read, const void *write, size_t size);

  bool forward_progress_needed () const { return m_forward_progress_needed; }
  void set_forward_progress_needed (bool forward_progress_needed);

  os_wave_launch_mode_t wave_launch_mode () const
  {
    return m_wave_launch_mode;
  }
  void set_wave_launch_mode (os_wave_launch_mode_t wave_launch_mode);

  void set_wave_launch_trap_override (os_wave_launch_trap_mask_t value,
                                      os_wave_launch_trap_mask_t mask);

  void set_precise_memory (bool enabled);

  /* Suspend/resume a list of queues.  Queues may become invalid as a result of
     suspension/resumption, but not destroyed.  Queues made invalid will
     destroy associated dispatches and waves.  Since waves/dispatches can be
     destroyed, the caller is responsible for refetching wave/dispatche
     instances from their id.  */
  size_t suspend_queues (const std::vector<queue_t *> &queues,
                         const char *reason) const;
  size_t resume_queues (const std::vector<queue_t *> &queues,
                        const char *reason) const;

  /* update_* ensures that the only objects that exist are exactly those
     reported by the os_driver.  It creates new objects reported by os_driver,
     and destroy objects that no longer exist which includes objects that are
     no longer valid.  When an object is deleted, all objects associated with
     it are also deleted.  */
  void update_agents ();
  void update_waves ();
  void update_queues ();
  void update_code_objects ();

  void runtime_enable (os_runtime_info_t runtime_info);

  void send_exceptions (
    os_exception_mask_t exceptions,
    std::variant<process_t *, agent_t *, queue_t *> source) const;

  void attach ();
  void detach ();

  void enqueue_event (event_t &event);
  event_t *next_pending_event ();

  size_t watchpoint_count () const;
  amd_dbgapi_watchpoint_share_kind_t watchpoint_shared_kind () const;
  void insert_watchpoint (const watchpoint_t &watchpoint);
  void remove_watchpoint (const watchpoint_t &watchpoint);
  const watchpoint_t *find_watchpoint (os_watch_id_t os_watch_id) const;

  static process_t &
  create_process (amd_dbgapi_client_process_id_t client_process_id)
  {
    return s_process_map.create_object (client_process_id);
  }
  static void destroy_process (process_t *process)
  {
    dbgapi_assert (process);
    s_process_map.destroy (process);
  }

  static auto all () { return s_process_map.range (); }
  static std::vector<process_t *> match (amd_dbgapi_process_id_t process_id);
  static process_t *find (amd_dbgapi_process_id_t process_id);
  static process_t *find (amd_dbgapi_client_process_id_t client_process_id);

  void get_info (amd_dbgapi_process_info_t query, size_t value_size,
                 void *value) const;

  amd_dbgapi_status_t get_os_pid (amd_dbgapi_os_process_id_t *pid) const;
  amd_dbgapi_status_t
  insert_breakpoint (amd_dbgapi_global_address_t address,
                     amd_dbgapi_breakpoint_id_t breakpoint_id);
  amd_dbgapi_status_t
  remove_breakpoint (amd_dbgapi_breakpoint_id_t breakpoint_id);

  template <typename Object, typename... Args> auto &create (Args &&...args)
  {
    return get_base_type_element<Object> (m_handle_object_sets)
      .template create_object<Object> (std::forward<Args> (args)...);
  }

  /* Destroy the given object.  */
  template <typename Object> void destroy (Object *object)
  {
    std::get<handle_object_set_t<Object>> (m_handle_object_sets)
      .destroy (object);
  }

  /* Destroy the object at object_it.  This should be used instead of
     process::destroy (Object *) if destroying objects while iterating
     with a process::range<Object>.  */
  template <typename ObjectIterator>
  ObjectIterator destroy (ObjectIterator object_it)
  {
    return std::get<handle_object_set_t<typename ObjectIterator::value_type>> (
             m_handle_object_sets)
      .destroy (object_it);
  }

  /* Return an Object range. A range implements begin () and end (), and
     can be used to iterate the Objects.  */
  template <typename Object> auto range ()
  {
    return std::get<handle_object_set_t<Object>> (m_handle_object_sets)
      .range ();
  }
  template <typename Object> auto range () const
  {
    return std::get<handle_object_set_t<Object>> (m_handle_object_sets)
      .range ();
  }

  /* Return the element count for the sub-Object.  */
  template <typename Object> size_t count () const
  {
    return std::get<handle_object_set_t<Object>> (m_handle_object_sets)
      .size ();
  }

  template <typename Object> bool changed () const
  {
    return std::get<handle_object_set_t<Object>> (m_handle_object_sets)
      .changed ();
  }

  /* Set the flag that indicates whether the Objects have changed. Return its
     previous value. The flag is set whenever objects are created, or
     destroyed, or invalidated.  */
  template <typename Object> bool set_changed (bool has_changed)
  {
    return std::get<handle_object_set_t<Object>> (m_handle_object_sets)
      .set_changed (has_changed);
  }

  /* Find an object with the given handle.  */
  template <typename Handle,
            std::enable_if_t<!std::is_void_v<object_type_from_handle_t<
                               Handle, decltype (m_handle_object_sets)>>,
                             int> = 0>
  auto *find (Handle id, bool all = false)
  {
    using object_type
      = object_type_from_handle_t<Handle, decltype (m_handle_object_sets)>;

    return std::get<handle_object_set_t<object_type>> (m_handle_object_sets)
      .find (id, all);
  }

  template <typename Handle,
            std::enable_if_t<!std::is_void_v<object_type_from_handle_t<
                               Handle, decltype (m_handle_object_sets)>>,
                             int> = 0>
  auto const *find (Handle id, bool all = false) const
  {
    using object_type
      = object_type_from_handle_t<Handle, decltype (m_handle_object_sets)>;

    return std::get<handle_object_set_t<object_type>> (m_handle_object_sets)
      .find (id, all);
  }

  /* Find an object for which the unary predicate f returns true.  */
  template <typename Functor>
  auto *find_if (Functor &&predicate, bool all = false)
  {
    using object_type = std::decay_t<utils::first_argument_of_t<Functor>>;

    return std::get<handle_object_set_t<object_type>> (m_handle_object_sets)
      .find_if (std::forward<Functor> (predicate), all);
  }

  template <typename Functor>
  auto const *find_if (Functor &&predicate, bool all = false) const
  {
    using object_type = std::decay_t<utils::first_argument_of_t<Functor>>;

    return std::get<handle_object_set_t<object_type>> (m_handle_object_sets)
      .find_if (std::forward<Functor> (predicate), all);
  }

  pipe_t &client_notifier_pipe () { return m_client_notifier_pipe; }
};

template <typename T>
void
process_t::read_global_memory (amd_dbgapi_global_address_t address, T *ptr,
                               size_t size)
{
  try
    {
      if (size_t xfer_size = read_global_memory_partial (address, ptr, size);
          xfer_size != size)
        throw memory_access_error_t (address + xfer_size);
    }
  catch (const memory_access_error_t &e)
    {
      fatal_error ("process_t::read_global_memory failed: %s", e.what ());
    }
}

template <typename T>
void
process_t::write_global_memory (amd_dbgapi_global_address_t address,
                                const T *ptr, size_t size)
{
  try
    {
      if (size_t xfer_size = write_global_memory_partial (address, ptr, size);
          xfer_size != size)
        throw memory_access_error_t (address + xfer_size);
    }
  catch (const memory_access_error_t &e)
    {
      fatal_error ("process_t::write_global_memory failed: %s", e.what ());
    }
}

namespace detail
{
template <typename Handle>
using process_find_t
  = decltype (std::declval<process_t> ().find (std::declval<Handle> ()));
} /* namespace detail */

/* Find an object with the given handle.  */
template <typename Handle,
          std::enable_if_t<
            utils::is_detected_v<detail::process_find_t, Handle>, int> = 0>
auto
find (Handle id) -> decltype (std::declval<process_t> ().find (id))
{
  if (detail::last_found_process)
    if (auto value = detail::last_found_process->find (id); value)
      return value;

  for (auto &&process : process_t::all ())
    {
      if (&process == detail::last_found_process)
        continue;

      if (auto value = process.find (id); value)
        {
          detail::last_found_process = &process;
          return value;
        }
    }

  return nullptr;
}

template <> struct is_flag<process_t::flag_t> : std::true_type
{
};

inline void
process_t::set_flag (flag_t flags)
{
  m_flags |= flags;
}

inline void
process_t::clear_flag (flag_t flags)
{
  m_flags &= ~flags;
}

inline bool
process_t::is_flag_set (flag_t flag) const
{
  dbgapi_assert (
    utils::is_power_of_two (static_cast<std::underlying_type_t<flag_t>> (flag))
    && "can only check one flag at a time");
  return (m_flags & flag) != 0;
}

namespace /* anonymous */
{

void
deallocate_memory (void *data)
{
  TRACE_CALLBACK_BEGIN (data);
  detail::process_callbacks.deallocate_memory (data);
  TRACE_CALLBACK_END ();
}

} /* namespace anonymous */

namespace detail
{

struct DeallocateMemory
{
  constexpr DeallocateMemory () noexcept = default;
  void operator() (void *mem) const noexcept { deallocate_memory (mem); }
};

} /* namespace detail */

template <typename T = void, typename D = detail::DeallocateMemory>
std::unique_ptr<T, D>
allocate_memory (size_t byte_size)
{
  TRACE_CALLBACK_BEGIN (byte_size);
  void *memory = detail::process_callbacks.allocate_memory (byte_size);

  if (!memory && byte_size)
    throw api_error_t (AMD_DBGAPI_STATUS_ERROR_CLIENT_CALLBACK);

  return std::unique_ptr<T, D> (
    static_cast<typename std::unique_ptr<T, D>::pointer> (memory));
  TRACE_CALLBACK_END ();
}

} /* namespace amd::dbgapi */

#endif /* AMD_DBGAPI_PROCESS_H */