qpcpp/ports/win32/qf_port.h
Quantum Leaps 75ef6eea18 5.6.4b
2016-05-05 12:22:15 -04:00

250 lines
9.1 KiB
C++

/// \file
/// \brief QF/C++ port to Win32 API
/// \cond
///***************************************************************************
/// Last updated for version 5.6.4
/// Last updated on 2016-05-04
///
/// Q u a n t u m L e a P s
/// ---------------------------
/// innovating embedded systems
///
/// Copyright (C) Quantum Leaps, LLC. All rights reserved.
///
/// This program is open source 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 3 of the License, or
/// (at your option) any later version.
///
/// Alternatively, this program may be distributed and modified under the
/// terms of Quantum Leaps commercial licenses, which expressly supersede
/// the GNU General Public License and are specifically designed for
/// licensees interested in retaining the proprietary status of their code.
///
/// This program is distributed in the hope that it will be useful,
/// but WITHOUT ANY WARRANTY; without even the implied warranty of
/// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
/// GNU General Public License for more details.
///
/// You should have received a copy of the GNU General Public License
/// along with this program. If not, see <http://www.gnu.org/licenses/>.
///
/// Contact information:
/// http://www.state-machine.com
/// mailto:info@state-machine.com
///***************************************************************************
/// \endcond
#ifndef qf_port_h
#define qf_port_h
// Win32 event queue and thread types
#define QF_EQUEUE_TYPE QEQueue
#define QF_OS_OBJECT_TYPE void*
#define QF_THREAD_TYPE void*
// The maximum number of active objects in the application
#define QF_MAX_ACTIVE 63
// The number of system clock tick rates
#define QF_MAX_TICK_RATE 2
// various QF object sizes configuration for this port
#define QF_EVENT_SIZ_SIZE 4
#define QF_EQUEUE_CTR_SIZE 4
#define QF_MPOOL_SIZ_SIZE 4
#define QF_MPOOL_CTR_SIZE 4
#define QF_TIMEEVT_CTR_SIZE 4
// QF interrupt disable/enable, see NOTE1
#define QF_INT_DISABLE() (QP::QF_enterCriticalSection_())
#define QF_INT_ENABLE() (QP::QF_leaveCriticalSection_())
// Win32 critical section
// QF_CRIT_STAT_TYPE -- not defined
#define QF_CRIT_ENTRY(dummy) QF_INT_DISABLE()
#define QF_CRIT_EXIT(dummy) QF_INT_ENABLE()
#include "qep_port.h" // QEP port
#include "qequeue.h" // Win32 needs event-queue
#include "qmpool.h" // Win32 needs memory-pool
#include "qpset.h" // Win32 needs priority-set
#include "qf.h" // QF platform-independent public interface
namespace QP {
void QF_enterCriticalSection_(void);
void QF_leaveCriticalSection_(void);
// set Win32 thread priority;
// can be called either before or after QMActive::START()
//
void QF_setWin32Prio(QMActive *act, int_t win32Prio);
// set clock tick rate
void QF_setTickRate(uint32_t ticksPerSec);
// clock tick callback (provided in the app)
void QF_onClockTick(void);
} // namespace QP
// special adaptations for QWIN GUI applications
#ifdef QWIN_GUI
// replace main() with main_gui() as the entry point to a GUI app.
#define main() main_gui()
int_t main_gui(); // prototype of the GUI application entry point
#endif
// portable "safe" facilities from <stdio.h> and <string.h> ...
#ifdef _MSC_VER /* Microsoft C/C++ compiler? */
#define SNPRINTF_S(buf_, len_, format_, ...) \
_snprintf_s(buf_, len_, _TRUNCATE, format_, ##__VA_ARGS__)
#define STRNCPY_S(dest_, src_, len_) \
strncpy_s(dest_, len_, src_, _TRUNCATE)
#define FOPEN_S(fp_, fName_, mode_) \
if (fopen_s(&fp_, fName_, mode_) != 0) { \
fp_ = static_cast<FILE *>(0); \
} else (void)0
#define CTIME_S(buf_, len_, time_) \
ctime_s((char *)buf_, len_, time_)
#define SSCANF_S(buf_, format_, ...) \
sscanf_s(buf_, format_, ##__VA_ARGS__)
#else // other C/C++ compilers (GNU, etc.)
#define SNPRINTF_S(buf_, len_, format_, ...) \
snprintf(buf_, len_, format_, ##__VA_ARGS__)
#define STRNCPY_S(dest_, src_, len_) strncpy(dest_, src_, len_)
#define FOPEN_S(fp_, fName_, mode_) \
(fp_ = fopen(fName_, mode_))
#define CTIME_S(buf_, len_, time_) \
strncpy(static_cast<char *>(buf_), ctime(time_), len_)
#define SSCANF_S(buf_, format_, ...) \
sscanf(buf_, format_, ##__VA_ARGS__)
#endif // _MSC_VER
//****************************************************************************
// interface used only inside QF, but not in applications
#ifdef QP_IMPL
// Win32-specific scheduler locking, see NOTE2
#define QF_SCHED_STAT_TYPE_ struct { uint_fast8_t m_lockPrio; }
#define QF_SCHED_LOCK_(pLockStat_, dummy) do { \
QF_enterCriticalSection_(); \
((pLockStat_)->m_lockPrio = \
static_cast<uint_fast8_t>(QF_MAX_ACTIVE)); \
} while (false)
#define QF_SCHED_UNLOCK_(dummy) QF_leaveCriticalSection_()
// Win32-specific event queue customization
#define QACTIVE_EQUEUE_WAIT_(me_) \
while ((me_)->m_eQueue.m_frontEvt == static_cast<QEvt const *>(0)) { \
QF_CRIT_EXIT_(); \
(void)WaitForSingleObject((me_)->m_osObject, (DWORD)INFINITE); \
QF_CRIT_ENTRY_(); \
}
#define QACTIVE_EQUEUE_SIGNAL_(me_) \
(void)SetEvent((me_)->m_osObject)
#define QACTIVE_EQUEUE_ONEMPTY_(me_) ((void)0)
// Win32-specific event pool operations
#define QF_EPOOL_TYPE_ QMPool
#define QF_EPOOL_INIT_(p_, poolSto_, poolSize_, evtSize_) do { \
uint_fast32_t fudgedSize = (poolSize_) * QF_WIN32_FUDGE_FACTOR; \
uint8_t *fudgedSto = new uint8_t[(poolSize_)*QF_WIN32_FUDGE_FACTOR]; \
Q_ASSERT_ID(210, fudgedSto != (uint8_t *)0); \
(void)(poolSto_); \
(p_).init(fudgedSto, fudgedSize, evtSize_); \
} while (false)
#define QF_EPOOL_EVENT_SIZE_(p_) ((p_).getBlockSize())
#define QF_EPOOL_GET_(p_, e_, m_) \
((e_) = static_cast<QEvt *>((p_).get((m_))))
#define QF_EPOOL_PUT_(p_, e_) ((p_).put(e_))
#define WIN32_LEAN_AND_MEAN
#include <windows.h> // Win32 API
namespace QP {
// Windows "fudge factor" for oversizing the resources, see NOTE2
enum {
QF_WIN32_FUDGE_FACTOR = 100
};
} // namespace QP
#endif // QP_IMPL
// NOTES: ====================================================================
//
// NOTE1:
// QF, like all real-time frameworks, needs to execute certain sections of
// code indivisibly to avoid data corruption. The most straightforward way of
// protecting such critical sections of code is disabling and enabling
// interrupts, which Win32 does not allow.
//
// This QF port uses therefore a single package-scope Win32 critical section
// object QF_win32CritSect_ to protect all critical sections.
//
// Using the single critical section object for all crtical section guarantees
// that only one thread at a time can execute inside a critical section. This
// prevents race conditions and data corruption.
//
// Please note, however, that the Win32 critical section implementation
// behaves differently than interrupt locking. A common Win32 critical section
// ensures that only one thread at a time can execute a critical section, but
// it does not guarantee that a context switch cannot occur within the
// critical section. In fact, such context switches probably will happen, but
// they should not cause concurrency hazards because the critical section
// eliminates all race conditionis.
//
// Unlinke simply disabling and enabling interrupts, the critical section
// approach is also subject to priority inversions. Various versions of
// Windows handle priority inversions differently, but it seems that most of
// them recognize priority inversions and dynamically adjust the priorities of
// threads to prevent it. Please refer to the MSN articles for more
// information.
//
// NOTE2:
// Scheduler locking (used inside QF_publish_()) is implemented in this
// port with the main critical section. This means that event multicasting
// will appear atomic, in the sense that no thread will be able to post
// events during multicasting.
//
// NOTE3:
// Windows is not a deterministic real-time system, which means that the
// system can occasionally and unexpectedly "choke and freeze" for a number
// of seconds. The designers of Windows have dealt with these sort of issues
// by massively oversizing the resources available to the applications. For
// example, the default Windows GUI message queues size is 10,000 entries,
// which can dynamically grow to an even larger number. Also the stacks of
// Win32 threads can dynamically grow to several megabytes.
//
// In contrast, the event queues, event pools, and stack size inside the
// real-time embedded (RTE) systems can be (and must be) much smaller,
// because you typically can put an upper bound on the real-time behavior
// and the resulting delays.
//
// To be able to run the unmodified applications designed originally for
// RTE systems on Windows, and to reduce the odds of resource shortages in
// this case, the generous QF_WIN32_FUDGE_FACTOR is used to oversize the
// event queues and event pools.
//
#endif // qf_port_h