rockbox/utils/hwstub/include/hwstub.hpp
Amaury Pouly e2271ed75e qeditor: fix compile error and some comments
Change-Id: I157c83fea8173adc53254f15aa49e41ee1ba7549
2017-08-28 22:52:29 +02:00

370 lines
16 KiB
C++

/***************************************************************************
* __________ __ ___.
* Open \______ \ ____ ____ | | _\_ |__ _______ ___
* Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
* Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
* Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
* \/ \/ \/ \/ \/
* $Id$
*
* Copyright (C) 2015 by Amaury Pouly
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
*
****************************************************************************/
#ifndef __HWSTUB_HPP__
#define __HWSTUB_HPP__
#include "hwstub_protocol.h"
#include <string>
#include <mutex>
#include <vector>
#include <cstdint>
#include <atomic>
#include <memory>
#include <chrono>
#include <thread>
#include <mutex>
#include <condition_variable>
#include <ostream>
#include <functional>
namespace hwstub {
class context;
class device;
class handle;
class context_poller;
/** C++ equivalent of /dev/null for streams */
extern std::ostream cnull;
extern std::wostream wcnull;
/** Errors */
enum class error
{
SUCCESS, /** Success */
ERROR, /** Unspecified error */
DISCONNECTED, /** Device has been disconnected */
PROBE_FAILURE, /** Device did not pass probing */
NO_CONTEXT, /** The context has been destroyed */
USB_ERROR, /** Unspecified USB error */
DUMMY, /** Call on dummy device/handle */
NO_SERVER, /** The server could not be reached */
SERVER_DISCONNECTED, /** Context got disconnected from the server */
SERVER_MISMATCH, /** The server is not compatible with hwstub */
PROTOCOL_ERROR, /** Network protocol error */
NET_ERROR, /** Network error */
TIMEOUT, /** Operation timed out */
OVERFLW, /** Operation stopped to prevent buffer overflow */
UNIMPLEMENTED, /** Operation has not been implemented */
UNSUPPORTED, /** Operation is not supported by device/protocol */
};
/** Return a string explaining the error */
std::string error_string(error err);
/** NOTE Multithreading:
* Unless specified, all methods are thread-safe
*/
/** Context
*
* A context provides a list of available devices and may notify devices
* arrival and departure.
*
* A context provides a way to regularly poll for derive changes. There are two
* ways to manually force an update:
* - on call to get_device_list(), the list is always refetched
* - on call to update_list() to force list update
* Note that automatic polling is disabled by default.
*/
class context : public std::enable_shared_from_this<context>
{
protected:
context();
public:
/** On destruction, the context will destroy all the devices. */
virtual ~context();
/** Get device list, clears the list in argument first. All devices in the list
* are still connected (or believe to be). This function will update the device
* list. */
error get_device_list(std::vector<std::shared_ptr<device>>& list);
/** Force the context to update its internal list of devices. */
error update_list();
/** Ask the context to automatically poll for device changes.
* Note that this might spawn a new thread to do so, in which case it will
* be destroyed/stop on deletetion or when stop_polling() is called. If
* polling is already enabled, this function will change the polling interval. */
void start_polling(std::chrono::milliseconds interval = std::chrono::milliseconds(250));
/** Stop polling. */
void stop_polling();
/** Register a notification callback with arguments (context,arrived,device)
* WARNING the callback may be called asynchronously ! */
typedef std::function<void(std::shared_ptr<context>, bool, std::shared_ptr<device>)> notification_callback_t;
typedef size_t callback_ref_t;
callback_ref_t register_callback(const notification_callback_t& fn);
void unregister_callback(callback_ref_t ref);
/** Return a dummy device that does nothing. A dummy device might be useful
* in cases where one still wants a valid pointer to no device. This dummy
* device does not appear in the list, it can be opened and will fail all requests. */
error get_dummy_device(std::shared_ptr<device>& dev);
/** Set/clear debug output for this context */
void set_debug(std::ostream& os);
inline void clear_debug() { set_debug(cnull); }
/** Get debug output for this context */
std::ostream& debug();
protected:
/** Notify the context about a device. If arrived is true, the device is
* added to the list and a reference will be added to it. If arrived is false,
* the device is marked as disconnected(), removed from the list and a
* reference will be removed from it. Adding a device that matches an
* existing one will do nothing. */
void change_device(bool arrived, std::shared_ptr<device> dev);
/** Do device notification */
void notify_device(bool arrived, std::shared_ptr<device> dev);
/** Opaque device type */
typedef void* ctx_dev_t;
/** Fetch the device list. Each item in the list is an opaque pointer. The function
* can also provide a pointer that will be used to free the list resources
* if necessary. Return <0 on error. */
virtual error fetch_device_list(std::vector<ctx_dev_t>& list, void*& ptr) = 0;
/** Destroy the resources created to get the list. */
virtual void destroy_device_list(void *ptr) = 0;
/** Create a new hwstub device from the opaque pointer. Return <0 on error.
* This function needs not add a reference to the newly created device. */
virtual error create_device(ctx_dev_t dev, std::shared_ptr<device>& hwdev) = 0;
/** Return true if the opaque pointer corresponds to the device. Only called
* from map_device(). */
virtual bool match_device(ctx_dev_t dev, std::shared_ptr<device> hwdev) = 0;
/** Check if a device matches another one in the list */
bool contains_dev(const std::vector<device*>& list, ctx_dev_t dev);
struct callback_t
{
notification_callback_t callback;
callback_ref_t ref;
};
std::shared_ptr<context_poller> m_poller; /* poller object */
std::recursive_mutex m_mutex; /* list mutex */
std::vector<std::shared_ptr<device>> m_devlist; /* list of devices */
std::vector<callback_t> m_callbacks; /* list of callbacks */
callback_ref_t m_next_cb_ref; /* next callback reference */
std::ostream *m_debug; /* debug stream */
};
/** Context Poller
*
* This class provides a way to regularly poll a context for device changes.
* NOTE this class is not meant to be used directly since context already
* provides access to it via start_polling() and stop_polling() */
class context_poller
{
public:
context_poller(std::weak_ptr<context> ctx, std::chrono::milliseconds interval = std::chrono::milliseconds(250));
~context_poller();
/** Set polling interval (in milliseconds) (works even if polling already enabled) */
void set_interval(std::chrono::milliseconds interval);
/** Start polling */
void start();
/** Stop polling. After return, no function will be made. */
void stop();
protected:
static void thread(context_poller *poller);
void poll();
std::weak_ptr<context> m_ctx; /* context */
bool m_running; /* are we running ? */
bool m_exit; /* exit flag for the thread */
std::thread m_thread; /* polling thread */
std::mutex m_mutex; /* mutex lock */
std::condition_variable m_cond; /* signalling condition */
std::chrono::milliseconds m_interval; /* Interval */
};
/** Device
*
* A device belongs to a context.
* Note that a device only keeps a weak pointer to the context, so it is possible
* for the context to be destroyed during the life of the device, in which case
* all operations on it will fail. */
class device : public std::enable_shared_from_this<device>
{
protected:
device(std::shared_ptr<context> ctx);
public:
virtual ~device();
/** Open a handle to the device. Several handles may be opened concurrently. */
error open(std::shared_ptr<handle>& handle);
/** Disconnect the device. This will notify the context that the device is gone. */
void disconnect();
/** Returns true if the device is still connected. */
bool connected();
/** Get context (might be empty) */
std::shared_ptr<context> get_context();
protected:
/** Some subsystems allow for hardware to be open several times and some do not.
* For example, libusb only allows one handle per device. To workaround this issue,
* open() will do some magic to allow for several open() even when the hardware
* supports only one. If the device does not support multiple
* handles (as reported by has_multiple_open()), open() will only call open_dev()
* the first time the device is opened and will redirect other open() calls to
* this handle using proxy handles. If the device supports multiple handles,
* open() will simply call open_dev() each time.
* The open_dev() does not need to care about this magic and only needs to
* open the device and returns the handle to it.
* NOTE this function is always called with the mutex locked already. */
virtual error open_dev(std::shared_ptr<handle>& handle) = 0;
/** Return true if device can be opened multiple times. In this case, each
* call to open() will generate a call to do_open(). Otherwise, proxy handles
* will be created for each open() and do_open() will only be called the first
* time. */
virtual bool has_multiple_open() const = 0;
std::weak_ptr<context> m_ctx; /* pointer to context */
std::recursive_mutex m_mutex; /* device state mutex: ref count, connection status */
bool m_connected; /* false once device is disconnected */
std::weak_ptr<handle> m_handle; /* weak pointer to the opened handle (if !has_multiple_open()) */
};
/** Handle
*
* A handle is tied to a device and provides access to the stub operation.
* The handle is reference counted and is destroyed
* when its reference count decreased to zero.
*/
class handle : public std::enable_shared_from_this<handle>
{
protected:
/** A handle will always hold a reference to the device */
handle(std::shared_ptr<device> dev);
public:
/** When destroyed, the handle will release its reference to the device */
virtual ~handle();
/** Return associated device */
std::shared_ptr<device> get_device();
/** Fetch a descriptor, buf_sz is the size of the buffer and is updated to
* reflect the number of bytes written to the buffer. */
error get_desc(uint16_t desc, void *buf, size_t& buf_sz);
/** Fetch part of the log, buf_sz is the size of the buffer and is updated to
* reflect the number of bytes written to the buffer. */
error get_log(void *buf, size_t& buf_sz);
/** Ask the stub to execute some code.
* NOTE: this may kill the stub */
error exec(uint32_t addr, uint16_t flags);
/** Read/write some device memory. sz is the size of the buffer and is updated to
* reflect the number of bytes written to the buffer.
* NOTE: the stub may or may not recover from bad read/write, so this may kill it.
* NOTE: the default implemtentation of read() and write() will split transfers
* according to the buffer size and call read_dev() and write_dev() */
error read(uint32_t addr, void *buf, size_t& sz, bool atomic);
error write(uint32_t addr, const void *buf, size_t& sz, bool atomic);
/** Execute a general cop operation: if out_data is not null, data is appended to header,
* if in_data is not null, a read operation follows to retrieve some data.
* The in_data parameters is updated to reflect the number of transfered bytes */
error cop_op(uint8_t op, uint8_t args[HWSTUB_COP_ARGS], const void *out_data,
size_t out_size, void *in_data, size_t *in_size);
/** Execute a coprocessor read operation */
error read32_cop(uint8_t args[HWSTUB_COP_ARGS], uint32_t& value);
/** Execute a coprocessor write operation */
error write32_cop(uint8_t args[HWSTUB_COP_ARGS], uint32_t value);
/** Get device buffer size: any read() or write() greater than this size
* will be split into several transaction to avoid overflowing device
* buffer. */
virtual size_t get_buffer_size() = 0;
/** Check a handle status. A successful handle does not guarantee successful
* operations. An invalid handle will typically report probing failure (the
* device did not pass probing) or disconnection.
* The default implemtentation will test if context still exists and connection status. */
virtual error status() const;
/** Shorthand for status() == HWSTUB_SUCCESS */
inline bool valid() const { return status() == error::SUCCESS; }
/** Helper functions */
error get_version_desc(hwstub_version_desc_t& desc);
error get_layout_desc(hwstub_layout_desc_t& desc);
error get_stmp_desc(hwstub_stmp_desc_t& desc);
error get_pp_desc(hwstub_pp_desc_t& desc);
error get_jz_desc(hwstub_jz_desc_t& desc);
error get_target_desc(hwstub_target_desc_t& desc);
protected:
/** The get_desc(), get_log(), exec(), read() and write() function
* take care of details so that each implementation can safely assume that
* the hwstub context exists and will not be destroyed during the execution
* of the function. It will also return early if the device has been disconnected.
*
* NOTE on read() and write():
* Since devices have a limited buffer, big transfers must be split into
* smaller ones. The high-level read() and write() functions perform this
* splitting in a generic way, based on get_buffer_size(), and calling read_dev()
* and write_dev() which do the actual operation.
* These function can safely assume that sz <= get_buffer_size(). */
virtual error read_dev(uint32_t addr, void *buf, size_t& sz, bool atomic) = 0;
virtual error write_dev(uint32_t addr, const void *buf, size_t& sz, bool atomic) = 0;
virtual error get_dev_desc(uint16_t desc, void *buf, size_t& buf_sz) = 0;
virtual error get_dev_log(void *buf, size_t& buf_sz) = 0;
virtual error exec_dev(uint32_t addr, uint16_t flags) = 0;
/* cop operation: if out_data is not null, data is appended to header, if
* in_data is not null, a READ2 operation follows to retrieve some data
* The in_data parameters is updated to reflect the number of transfered bytes*/
virtual error cop_dev(uint8_t op, uint8_t args[HWSTUB_COP_ARGS], const void *out_data,
size_t out_size, void *in_data, size_t *in_size) = 0;
std::shared_ptr<device> m_dev; /* pointer to device */
std::atomic<int> m_refcnt; /* reference count */
std::recursive_mutex m_mutex; /* operation mutex to serialise operations */
};
/** Dummy device */
class dummy_device : public device
{
friend class context; /* for ctor */
protected:
dummy_device(std::shared_ptr<context> ctx);
public:
virtual ~dummy_device();
protected:
virtual error open_dev(std::shared_ptr<handle>& handle);
virtual bool has_multiple_open() const;
};
/** Dummy handle */
class dummy_handle : public handle
{
friend class dummy_device;
protected:
dummy_handle(std::shared_ptr<device> dev);
public:
virtual ~dummy_handle();
protected:
virtual error read_dev(uint32_t addr, void *buf, size_t& sz, bool atomic);
virtual error write_dev(uint32_t addr, const void *buf, size_t& sz, bool atomic);
virtual error get_dev_desc(uint16_t desc, void *buf, size_t& buf_sz);
virtual error get_dev_log(void *buf, size_t& buf_sz);
virtual error exec_dev(uint32_t addr, uint16_t flags);
virtual error cop_dev(uint8_t op, uint8_t args[HWSTUB_COP_ARGS], const void *out_data,
size_t out_size, void *in_data, size_t *in_size);
virtual error status() const;
virtual size_t get_buffer_size();
struct hwstub_version_desc_t m_desc_version;
struct hwstub_layout_desc_t m_desc_layout;
struct hwstub_target_desc_t m_desc_target;
};
} // namespace hwstub
#endif /* __HWSTUB_HPP__ */