CPUCore.cpp

Go to the documentation of this file.

 
#include "etiss/CPUCore.h"
#include "etiss/ETISS.h"
#include "etiss/CoreDSLCoverage.h"
 
using namespace etiss;
 
void CPUArchRegListenerInterface::signalChangedRegisterValue(ETISS_CPU *cpu, const char *registerName)
{
    CPUCore *core = (CPUCore *)cpu->_etiss_private_handle_;
    if (!core)
    {
        etiss::log(
            etiss::ERROR,
            "CPUArchRegListenerInterface::signalChangedRegisterValue() called from outside etiss::CPUCore::execute(). "
            "this should not happen and indicates a faultiy CPUArch (or Plugin) implementation. This function may have "
            "been called indirectly from ETISS_signalChangedRegisterValue()");
        return;
    }
    auto vs = core->getStruct();
    if (!vs)
    {
        etiss::log(etiss::ERROR, "CPUArchRegListenerInterface::signalChangedRegisterValue() called but CPUArch didn't "
                                 "provide a VirtualStruct.");
        return;
    }
    auto field = vs->findName(registerName);
    if (!field)
    {
        field = vs->findPrettyName(registerName);
        if (!field)
        {
            etiss::log(etiss::ERROR,
                       "CPUArchRegListenerInterface::signalChangedRegisterValue() called but the associated "
                       "VirtualStruct has not the specified field.",
                       registerName);
            return;
        }
    }
    if (!(field->flags_ & etiss::VirtualStruct::Field::L))
    {
        etiss::log(etiss::ERROR,
                   "CPUArchRegListenerInterface::signalChangedRegisterValue() called but the field of the associated "
                   "VirtualStruct doesn't have the listerner flags set (etiss::VirtualStruct::Field::L).",
                   registerName);
        return;
    }
 
    field->signalWrite();
}
 
CPUCore::InterruptVectorWrapper::InterruptVectorWrapper(CPUCore &parent) : parent_(parent) {}
void CPUCore::InterruptVectorWrapper::setBit(unsigned bit, bool state)
{
    bool consumed = false;
    // std::cout << "CPUCore::InterruptVectorWrapper::setBit called " << state << " " << bit << std::endl;
    // iterate through the plugins which have been registered at the cpu
    for (auto iter : parent_.plugins)
    {
        etiss::InterruptListenerPlugin *ilp = iter->getInterruptListenerPlugin();
        if (ilp)
            consumed |= ilp->interruptWrite(bit, state);
    }
    if (!consumed)
    {
        consumed_by_interruptlistener_ = false;
        // std::cout << "CPUCore::InterruptVectorWrapper::setBit NOT consumed by InterruptListenerPlugin" << std::endl;
        parent_.intvector_->setBit(bit, state);
    }
    else
    {
        consumed_by_interruptlistener_ = true;
        // std::cout << "CPUCore::InterruptVectorWrapper::setBit consumed by InterruptListenerPlugin" << std::endl;
    }
}
bool CPUCore::InterruptVectorWrapper::getBit(unsigned bit) const
{
    return parent_.intvector_->getBit(bit);
}
unsigned CPUCore::InterruptVectorWrapper::width() const
{
    return parent_.intvector_->width();
}
bool CPUCore::InterruptVectorWrapper::isActive() const
{
    return parent_.intvector_->isActive();
}
void CPUCore::InterruptVectorWrapper::clear()
{
    for (unsigned i = 0; i < width(); i++)
    {
        setBit(i, false);
    }
}
 
std::mutex CPUCore::instances_mu_;
std::list<std::weak_ptr<CPUCore>> CPUCore::instances_;
 
int currID = 0;
int CPUCore::getNextID()
{
    return currID;
}
CPUCore::CPUCore(std::shared_ptr<etiss::CPUArch> arch)
    : arch_(arch)
    , name_("core" + std::to_string(currID))
    , id_(currID++)
    , cpu_(arch->newCPU())
    , vcpu_(arch->getVirtualStruct(cpu_))
    , intvector_(arch->createInterruptVector(cpu_))
    , intenable_(arch->createInterruptEnable(cpu_))
    , mmu_enabled_(false)
{
    arch_->resetCPU(cpu_, 0);
    timer_enabled_ = true;
    bcc_ = 1;
    exception_skip_count_ = 0;
    blockCacheLimit_ = -1;
    intwrapper_ = intvector_ ? new InterruptVectorWrapper(*this) : 0;
    blockCounter = 0;
#if ETISS_CPUCORE_DBG_APPROXIMATE_INSTRUCTION_COUNTER
    instrcounter = 0;
#endif
 
    if (cpu_)
    {
        if (vcpu_)
        {
            if (!vcpu_->findName("instructionPointer"))
            {
                ETISS_CPU *cpu = cpu_;
                vcpu_->addField((new etiss::VirtualStruct::Field(
                                     *vcpu_, "instructionPointer", "",
                                     etiss::VirtualStruct::Field::R | etiss::VirtualStruct::Field::P, 8, false,
                                     [cpu]() { return (uint64_t)cpu->instructionPointer; }, nullptr))
                                    ->setDeleteP(std::function<void(etiss::VirtualStruct::Field *)>(
                                        [](etiss::VirtualStruct::Field *f) { delete f; })));
            }
#if ETISS_CPUCORE_DBG_APPROXIMATE_INSTRUCTION_COUNTER
            if (!vcpu_->findName("instructionCounter"))
            {
                vcpu_->addField((new etiss::VirtualStruct::Field(
                                     *vcpu_, "instructionCounter", "",
                                     etiss::VirtualStruct::Field::R | etiss::VirtualStruct::Field::W |
                                         etiss::VirtualStruct::Field::P,
                                     8, false, [this]() { return (uint64_t)instrcounter; },
                                     [this](uint64_t v) { instrcounter = v; }))
                                    ->setDeleteP(std::function<void(etiss::VirtualStruct::Field *)>(
                                        [](etiss::VirtualStruct::Field *f) { delete f; })));
            }
#endif
        }
    }
}
 
void CPUCore::addPlugin(std::shared_ptr<etiss::Plugin> plugin)
{
    if (plugin.get() != 0)
    {
        {
            std::lock_guard<std::mutex> lock(mu_);
            // check if the plugin is already present
            for (const std::shared_ptr<etiss::Plugin> &p : plugins)
            {
                if (p.get() == plugin.get())
                {
                    etiss::log(etiss::WARNING, "cannot add the same plugin multiple times", *this, *(plugin.get()));
                    return;
                }
            }
            if (plugin->plugin_core_)
            {
                etiss::log(etiss::WARNING, "A plugin has been added to multiple cores. This could be a serve problem.",
                           *this, *(plugin.get()));
            }
            plugins.push_back(plugin);
        }
        plugin->plugin_core_ = this;
        plugin->setCorrespondingCPUCoreName(this->getName());
        plugin->addedToCPUCore(this);
    }
    else
    {
        etiss::log(etiss::WARNING, "etiss::CPUCore::addPlugin() called without passing a valid plugin pointer.", *this);
    }
}
 
void CPUCore::removePlugin(std::shared_ptr<etiss::Plugin> plugin)
{
    etiss::Plugin *p = plugin.get();
    bool removed = false;
    {
        std::lock_guard<std::mutex> lock(mu_);
        for (auto iter = plugins.begin(); iter != plugins.end();)
        {
            if (iter->get() == plugin.get())
            {
                plugins.erase(iter++);
                removed = true;
                break;
            }
            else
            {
                iter++;
            }
        }
    }
    if (removed)
    {
        p->plugin_core_ = nullptr;
        p->removedFromCPUCore(this);
    }
}
 
std::shared_ptr<CPUCore> CPUCore::create(std::string archname, std::string instancename,
                                         std::map<std::string, std::string> archoptions)
{
 
    etiss::forceInitialization();
 
    // get arch -> constructs arch with used arch library
    std::shared_ptr<etiss::CPUArch> arch = getCPUArch(archname, archoptions);
 
    if (arch.get() == 0)
    {
        etiss::log(etiss::FATALERROR, "Architecture not found.", archname);
        return 0;
    }
 
    // creat core
    std::shared_ptr<CPUCore> ret(new CPUCore(arch));
 
    {
        std::lock_guard<std::mutex> lock(instances_mu_);
        instances_.emplace_back(ret);
    }
 
    return ret;
}
 
std::list<std::string> CPUCore::list()
{
    std::list<std::string> ret;
    {
        std::lock_guard<std::mutex> lock(instances_mu_);
        for (auto iter = instances_.begin(); iter != instances_.end(); iter++)
        {
            auto ptr = iter->lock();
            if (ptr.get() != 0)
            {
                ret.push_back(ptr->getName() + " [" + ptr->getArch()->getArchName() + "," + ptr->getJITName() +
                              "]"); // "CORENAME [ARCHITECTURE,JIT]
            }
        }
    }
    return ret;
}
 
CPUCore::~CPUCore()
{
    arch_->deleteInterruptVector(intvector_, cpu_);
    if (vcpu_)
    {
        vcpu_->close();
        vcpu_.reset();
    }
    arch_->deleteCPU(cpu_);
    delete intwrapper_;
}
 
static bool verifyJITSizeOf(std::string structname, etiss::int32 expected_size, etiss::JIT *jit,
                            std::string prefix = std::string())
{
    if (jit == 0)
        return false;
    // generate code
    std::string error;
    std::string code = std::string(prefix + "\n#include \"etiss/jit/CPU.h\"\n#include "
                                            "\"etiss/jit/System.h\"\n#include \"etiss/jit/ReturnCode.h\"\n "
                                            "#include \"etiss/jit/types.h\"\n#include "
                                            "\"etiss/jit/fpu/softfloat.h\"\n etiss_int32 get_size(){ return "
                                            "sizeof(") +
                       structname + ");}";
 
    std::set<std::string> headers;
    headers.insert(etiss::jitFiles());
    // compile
    void *handle = jit->translate(code, headers, std::set<std::string>(), std::set<std::string>(), error, true);
    if (handle == 0)
    {
        etiss::log(etiss::ERROR,
                   std::string("Failed to compile test code [") + code + "] to check struct size: " + error);
        return false;
    }
    // check size
    typedef etiss::int32 (*get_size)(void);
    get_size gs = (get_size)jit->getFunction(handle, "get_size", error);
    if (gs == 0)
    {
        jit->free(handle);
        etiss::log(etiss::ERROR,
                   std::string("Failed to get compiled function [get_size] to check struct size: ") + error);
        return false;
    }
    etiss::int32 r = gs();
    jit->free(handle);
    if (r != expected_size)
    {
        std::stringstream ss;
        ss << "Unexpected size of " << structname << ";";
        ss << " Expected: " << expected_size;
        ss << " Is: " << r;
        etiss::log(etiss::ERROR, ss.str());
        return false;
    }
    return true;
}
static bool verifyJITPragmaPack(etiss::JIT *jit)
{
    if (jit == 0)
        return false;
    std::string error;
    std::stringstream stru;
    etiss::int32 expected_size = 0;
 
    // add some types to test
    std::vector<const char *> types;
    std::vector<etiss::int32> typeslen;
    types.push_back("etiss_int8");
    typeslen.push_back(sizeof(etiss_int8));
    types.push_back("etiss_uint8");
    typeslen.push_back(sizeof(etiss_uint8));
    types.push_back("etiss_int16");
    typeslen.push_back(sizeof(etiss_int16));
    types.push_back("etiss_uint16");
    typeslen.push_back(sizeof(etiss_uint16));
    types.push_back("etiss_int32");
    typeslen.push_back(sizeof(etiss_int32));
    types.push_back("etiss_uint32");
    typeslen.push_back(sizeof(etiss_uint32));
    types.push_back("etiss_int64");
    typeslen.push_back(sizeof(etiss_int64));
    types.push_back("etiss_uint64");
    typeslen.push_back(sizeof(etiss_uint64));
    types.push_back("void*");
    typeslen.push_back(sizeof(void *));
 
    stru << "struct _etiss_test_struct {\n";
    // add eacht type followed by any other type to create different combinations
    for (size_t i = 0; i < types.size(); i++)
    {
        stru << types[i] << " var_" << i << ";\n";
        expected_size += typeslen[i];
        for (size_t j = 0; j < types.size(); j++)
        {
            if (j != i)
            {
                stru << types[j] << " var_" << i << "_" << j << ";\n";
                expected_size += typeslen[j];
            }
        }
    }
 
    stru << "};";
 
    std::string code = std::string("\n#include \"etiss/jit/types.h\"\n#pragma pack(push, 1)\n") + stru.str() +
                       "\n#pragma pack(pop)\n etiss_int32 get_size(){ return sizeof(struct "
                       "_etiss_test_struct);}";
 
    std::set<std::string> headers;
    headers.insert(etiss::jitFiles());
    // compile
    void *handle = jit->translate(code, headers, std::set<std::string>(), std::set<std::string>(), error, true);
    if (handle == 0)
    {
        etiss::log(etiss::ERROR,
                   std::string("Failed to compile test code [") + code + "] to check struct size: " + error);
        return false;
    }
    typedef etiss::int32 (*get_size)(void);
    // check size
    get_size gs = (get_size)jit->getFunction(handle, "get_size", error);
    if (gs == 0)
    {
        jit->free(handle);
        etiss::log(etiss::ERROR,
                   std::string("Failed to get compiled function [get_size] to check struct size: ") + error);
        return false;
    }
    etiss::int32 r = gs();
    jit->free(handle);
    if (r != expected_size)
    {
        std::stringstream ss;
        ss << "Unexpected size of test structure;";
        ss << " Expected: " << expected_size;
        ss << " Is: " << r;
        etiss::log(etiss::ERROR, ss.str());
        return false;
    }
    return true;
}
 
static void etiss_CPUCore_handleException(ETISS_CPU *cpu, etiss::int32 &code, BlockLink *&block_ptr,
                                          Translation &translator, CPUArch *arch)
{
 
#if DEBUG
    if (unlikely(code == RETURNCODE::NOERROR))
    {
        etiss::log(etiss::ERROR, "etiss_CPUCore_handleException may not be called with error code NOERROR");
    }
#endif
 
    switch (code)
    {
    case RETURNCODE::RELOADBLOCKS:
        block_ptr = 0; // doesn't hold a reference and thus might become invalid
        translator.unloadBlocks();
        code = RETURNCODE::NOERROR;
        return;
    case RETURNCODE::RELOADCURRENTBLOCK:
        if (block_ptr)
            block_ptr->valid = false; // invalidate but don't delete block
        block_ptr = 0;
        code = RETURNCODE::NOERROR;
        return;
    case RETURNCODE::GDBNOERROR:
        code = RETURNCODE::NOERROR;
        return;
    case RETURNCODE::CPUFINISHED:
        return;
    default:
        code = arch->handleException(code, cpu);
        return;
    }
}
 
class LegacyRegisterDevicePluginListener : public etiss::VirtualStruct::Field::Listener
{
  public:
    std::list<etiss::RegisterDevicePlugin *> plugins;
    LegacyRegisterDevicePluginListener(const std::list<etiss::RegisterDevicePlugin *> &plugins_) : plugins(plugins_) {}
    virtual ~LegacyRegisterDevicePluginListener() {}
    virtual void write(etiss::VirtualStruct::Field &field, uint64_t val)
    {
        std::string name = field.name_;
        const char *cname = name.c_str();
        for (auto plugin : plugins)
        {
            if (plugin)
            {
                plugin->changedRegister(cname);
            }
        }
    }
};
 
etiss::int32 CPUCore::execute(ETISS_System &_system)
{
    ETISS_System *system = &_system; // change to pointer for reassignments
 
    if (!ETISS_System_isvalid(system))
    { // check if required functions are present
        return RETURNCODE::INVALIDSYSTEM;
    }
 
    std::lock_guard<std::mutex> lock(mu_); // lock class fields from modification
 
    if (!arch_)
    {
        etiss::log(etiss::ERROR, "Could not find architecture!");
        return RETURNCODE::GENERALERROR;
    }
 
    if (!cpu_)
    {
        etiss::log(etiss::ERROR, "Could not find CPU struct!");
        return RETURNCODE::GENERALERROR;
    }
 
    cpu_->_etiss_private_handle_ =
        (void *)this; // init pointer to execute RegisterDevicePlugins. the value of tis pointer may be invalid/subject
                      // to change and may not be used by external code
 
    // get JIT instance
    std::shared_ptr<JIT> jiti = jit_; // copy jit because it may change
    if (!jiti)
    {
        etiss::log(etiss::ERROR, std::string("No JIT available to ") + name_);
        return RETURNCODE::JITERROR;
    }
 
 
    // verify jit
    if (etiss::cfg().get<bool>("jit.verify", true))
    {
        if (!verifyJITSizeOf("etiss_int64", sizeof(etiss_int64), jiti.get()))
            return RETURNCODE::JITCOMPILATIONERROR;
        if (!verifyJITSizeOf("etiss_int32", sizeof(etiss_int32), jiti.get()))
            return RETURNCODE::JITCOMPILATIONERROR;
        if (!verifyJITSizeOf("etiss_int16", sizeof(etiss_int16), jiti.get()))
            return RETURNCODE::JITCOMPILATIONERROR;
        if (!verifyJITSizeOf("etiss_int8", sizeof(etiss_int8), jiti.get()))
            return RETURNCODE::JITCOMPILATIONERROR;
        if (!verifyJITSizeOf("ETISS_CPU", sizeof(ETISS_CPU), jiti.get()))
            return RETURNCODE::JITCOMPILATIONERROR;
        if (!verifyJITPragmaPack(jiti.get()))
            return RETURNCODE::JITCOMPILATIONERROR;
        etiss::log(etiss::INFO, std::string("JIT compiler ") + jiti->getName() +
                                    " has passed the verification tests (tested by CPUCore " + name_ + ")");
    }
    // add default timer plugin from arch
    if (timer_enabled_)
    {
        Plugin *timerInstance = arch_->newTimer(cpu_);
        if (!timerInstance)
        {
            etiss::log(etiss::ERROR, "ERROR: default timer requested but not supported by architecture");
            return RETURNCODE::GENERALERROR;
        }
        else
        {
            etiss::log(etiss::INFO, "Add Timer Plugin: " + timerInstance->getPluginName());
            auto local_arch = arch_;
            plugins.push_back(std::shared_ptr<etiss::Plugin>(timerInstance, [local_arch](etiss::Plugin *p) {
                etiss::log(etiss::INFO, "Delete Timer Plugin.");
                local_arch->deleteTimer(p);
            }));
        }
    }
 
    // add MMU module from the arch
    {
        etiss::mm::MMU *new_mmu = arch_->newMMU(cpu_);
        if (new_mmu)
        {
            mmu_enabled_ = true;
            mmu_.reset(new_mmu);
            etiss::log(etiss::INFO, "Add MMU module: " + mmu_->GetName());
        }
    }
 
    if (mmu_enabled_)
    {
        plugins.push_back(std::make_shared<etiss::mm::DMMUWrapper>(mmu_));
    }
 
    // copy system wrapper plugins to list and update system (pre plugin init)
    std::list<SystemWrapperPlugin *> syswrappers;
    for (auto &plugin : plugins)
    {
        auto c = plugin->getSystemWrapperPlugin();
        if (c)
        {
            ETISS_System *wsys = c->wrap(cpu_, system);
            if (wsys)
            {
                syswrappers.push_front(c); // inverse order for easy iteration
                system = wsys;
            }
            else
            {
                std::stringstream stream;
                stream << "SystemWrapperPlugin \"" << c->getPluginName() << "\" failed to wrap ETISS_System instance";
                etiss::log(etiss::WARNING, stream.str());
            }
        }
    }
 
    // initialize plugins
    for (auto &p : plugins)
    {
        if (!p)
            etiss::log(etiss::FATALERROR, "Empty plugin");
 
        p->plugin_cpu_ = cpu_;
        p->plugin_system_ = system;
        p->plugin_arch_ = arch_.get();
        p->init(cpu_, system, arch_.get());
 
        std::stringstream m;
        m << "Init Plugin " << p->getPluginName();
        etiss::log(etiss::INFO, m.str());
    }
 
    // copy coroutine plugins to array
    std::vector<CoroutinePlugin *> cor_array;
    for (const auto &plugin : plugins)
    {
        auto c = plugin->getCoroutinePlugin();
        if (c)
            cor_array.push_back(c);
    }
 
    // create translation object
    Translation translation(arch_, jiti, plugins, *system, *cpu_);
 
    // Translation init returns a list of pluigins, at position 0 this is the arch plugin followed by all translation
    // plugins
    void **plugins_handle_ = translation.init();
    if (!plugins_handle_)
    {
        etiss::log(etiss::FATALERROR, "Failed to initialize translation");
    }
 
    // enable RegisterDevicePlugin listeneing by adding a listener to all fields of the VirtualStruct
    etiss::VirtualStruct::Field::Listener *listener = 0;
    {
        std::list<RegisterDevicePlugin *> regdevices;
        for (auto &plugin : plugins)
        {
            auto rdp = plugin->getRegisterDevicePlugin();
            if (rdp)
                regdevices.push_back(rdp);
        }
        if (!regdevices.empty())
        {
            etiss::log(
                etiss::INFO,
                "etiss::RegisterDevicePlugin is a legacy convenience plugin. it may become deprecated later on. "
                "consider using etiss::VirtualStruct::Field::Listener to directly listen only for relevant fields.");
 
            if (vcpu_)
            {
                listener = new LegacyRegisterDevicePluginListener(regdevices);
 
                vcpu_->foreachField([listener](std::shared_ptr<etiss::VirtualStruct::Field> f) {
                    f->addListener(listener);
                }); // add listener to all current field of struct
 
                // TODO: maybe later VirtualStruct will support a listener for added/removed fields. in that case the
                // lisener of this function should also be added to new fields
            }
            else
            {
                etiss::log(etiss::WARNING,
                           "etiss::RegisterDevicePlugin added to a CPUCore that doesn't have a VirtualStruct.", name_);
            }
        }
    }
 
    // start execution loop
 
    bool exit_on_loop = etiss::cfg().get<bool>("etiss.exit_on_loop", false);
 
    float startTime = (float)clock() / CLOCKS_PER_SEC; // TESTING
 
    BlockLink *blptr = 0; // pointer to the current block
 
    etiss::int32 exception = RETURNCODE::NOERROR;
 
    // sync time at the beginning (e.g. SystemC processes running at time 0)
    system->syncTime(system->handle, cpu_);
 
    // execution loop
    {
#if ETISS_DBG_ICOUNT_LIMIT > 0 && ETISS_CPUCORE_DBG_APPROXIMATE_INSTRUCTION_COUNTER
        while (likely(instrcounter < ETISS_DBG_ICOUNT_LIMIT))
        {
#else
        while (true)
        {
#endif
 
            // execute coroutines
            for (auto &cor_plugin : cor_array)
            {
                exception = cor_plugin->execute();
                if (unlikely(exception != RETURNCODE::NOERROR)) // check exception
                {
                    etiss_CPUCore_handleException(cpu_, exception, blptr, translation, arch_.get()); // handle exception
                    if (unlikely(exception != RETURNCODE::NOERROR)) // check if exception handling failed
                    {
                        goto loopexit; // return exception; terminate cpu
                    }
                }
            }
            //            std::cout << "blockCounter: " <<  blockCounter++ <<std::endl;
            //            std::cout << "instrcounter: " <<  instrcounter <<std::endl;
            for (unsigned bc = 0; bc < bcc_; bc++)
            {
                // if not block internal jump // NOTE: removed since tests showed that this decreases performance
                // if (!(blptr != 0 && blptr->valid && blptr->start<=cpu->instructionPointer && blptr->end >
                // cpu->instructionPointer)){
                // Transalte virtual address to physical address if MMU is enabled
                uint64_t pma = cpu_->instructionPointer;
 
                // remember pc and cpu time to check for loop to self instructions
                uint64_t old_pc = cpu_->instructionPointer;
                uint64_t old_time = cpu_->cpuTime_ps;
 
                if (mmu_enabled_)
                {
                    if (mmu_->cache_flush_pending)
                    {
                        // FIXME: When flush required, current instruction cache has to be cleared. However, the
                        // unloadBlocks is much too time-comsuming than expected. It should be optimized later on.
                        // translation.unloadBlocks(0,(uint64_t)((int64_t)-1));
                        mmu_->cache_flush_pending = false;
                        blptr = nullptr;
                    }
 
                    // If the exception could be handled by architecture, then continue translation
                    while ((exception = mmu_->Translate(cpu_->instructionPointer, &pma, etiss::mm::X_ACCESS)))
                    {
                        //      translation.unloadBlocks();
                        if ((exception = arch_->handleException(exception, cpu_)))
                            goto loopexit;
                        // Update pma, in case pc is redirected to physical address space
                        pma = cpu_->instructionPointer;
                    }
                }
 
                // FIXME: cpu->instructionPointer contains virtual address, getBlockFast should use physical address
                // instead to realize physical cache.
                blptr = translation.getBlockFast(
                    blptr, cpu_->instructionPointer); // IMPORTANT: no pointer reference is kept here. if the translator
                                                      // performs a cleanup then blptr must be set to 0
                //}
 
                if (unlikely(blptr == 0)) // if no block function pointer could be acquired
                {
                    if (false)
                    {
                        // emulation interface? switch jit?
                        exception = RETURNCODE::ARCHERROR;
                        goto loopexit;
                    }
                    else
                    {
                        std::stringstream stream;
                        stream << "CPU execution stopped: Cannot execute from instruction index " << std::hex
                               << cpu_->instructionPointer << std::dec << ": no translated code available" << std::endl;
                        etiss::log(etiss::WARNING, stream.str());
                        exception = RETURNCODE::JITCOMPILATIONERROR;
                        goto loopexit;
                    }
                }
                else
                {
                    // etiss::log(etiss::FATALERROR,"disabled etiss iss");
#if ETISS_CPUCORE_DBG_APPROXIMATE_INSTRUCTION_COUNTER
                    uint64 oldinstrptr = cpu_->instructionPointer; // TESTING
#endif
                    // plugins_handle_ has the pointer to all translation plugins,
                    // In the generated code these plugin handles are named "plugin_pointers" and can be used to access
                    // a variable of the plugin
                    exception = (*(blptr->execBlock))(cpu_, system, plugins_handle_);
 
                    // exit simulator when a loop to self instruction is encountered
                    if (exit_on_loop && !exception &&
                            old_time + cpu_->cpuCycleTime_ps == cpu_->cpuTime_ps &&
                            old_pc == cpu_->instructionPointer)
                    {
                        exception = RETURNCODE::CPUFINISHED;
                    }
 
#if ETISS_CPUCORE_DBG_APPROXIMATE_INSTRUCTION_COUNTER
                    instrcounter +=
                        blptr->end - oldinstrptr; // TESTING ///TODO handle early exception exit? ///BUG:
                                                  // InstructionPointer increases more than 1 per instruction!!!
//                    std::cout << "blocksize: " << etiss::toString(blptr->end-oldinstrptr) << std::endl;
#endif
                }
 
                // check for exception in executed block
                if (unlikely(exception != RETURNCODE::NOERROR))
                {
                    etiss_CPUCore_handleException(cpu_, exception, blptr, translation, arch_.get()); // handle exception
                    if (unlikely(exception != RETURNCODE::NOERROR)) // check if exception handling failed
                    {
                        goto loopexit; // exception; terminate cpu
                    }
                }
            }
 
            // sync time after block
            system->syncTime(system->handle, cpu_);
        }
    }
 
loopexit:
 
    float endTime = (float)clock() / CLOCKS_PER_SEC;
 
 
    // execute coroutines end
    for (auto &cor_plugin : cor_array)
    {
        cor_plugin->executionEnd(exception);
    }
 
    // Defining the statistics of measurement and printing them
    double cpu_time = cpu_->cpuTime_ps / 1.0E12;
    double simulation_time = endTime - startTime;
    double cpu_cycle = cpu_->cpuTime_ps / (float)cpu_->cpuCycleTime_ps;
    double mips = cpu_->cpuTime_ps / (float)cpu_->cpuCycleTime_ps / simulation_time / 1.0E6;
    std::cout << "CPU Time: " << (cpu_time) << "s    Simulation Time: " << (simulation_time) << "s"
              << std::endl;
    std::cout << "CPU Cycles (estimated): " << (cpu_cycle) << std::endl;
    std::cout << "MIPS (estimated): " << (mips) << std::endl;
 
 
    // declaring path of writing the json file contaiing performance metrics and the boolean which approves of writing the json output
    std::string valid_json_output_path = etiss::cfg().get<std::string>("vp.stats_file_path", "");
    bool output_json = etiss::cfg().isSet("vp.stats_file_path");
 
    if(output_json==true)
    {
        std::ofstream json_output(valid_json_output_path);
        json_output << "{\"mips\": " << mips << ", \"Simulation_Time\": " << simulation_time << ", \"CPU_Time\": " << cpu_time << ", \"CPU_cycle\": " << cpu_cycle << "}" << std::endl;
    }
 
    #ifndef ETISS_USE_COREDSL_COVERAGE
    if (etiss::cfg().isSet("vp.coredsl_coverage_path")) {
        etiss::log(etiss::WARNING, "Coverage Analysis is disabled but vp.coredsl_coverage_path is set. To enable coverage analysis, build ETISS with -DETISS_USE_COREDSL_COVERAGE");
    }
    #endif
 
    std::string coverage_output_path = etiss::cfg().get<std::string>("vp.coredsl_coverage_path", "coverage.csv");
    if (!coverage_map.empty()) {
        std::ofstream coverage_output(coverage_output_path);
        coverage_output << arch_->getArchName() << std::endl;
        coverage_output << "ID;Count" << std::endl;
        for (auto it : coverage_map) {
            coverage_output << it.first << ";" << it.second << std::endl;
        }
    }
 
 
    etiss_uint64 max = 0;
    for (int i = 0; i < ETISS_MAX_RESOURCES; i++)
    {
        if (cpu_->resources[i])
        {
            if (cpu_->cycles[i] > max)
            {
                max = cpu_->cycles[i];
            }
        }
    }
    if (max != 0)
    { // max=0: resource computation turned of
        std::cout << "CPU Cycles (with pipeline): " << max << std::endl;
    }
    for (int i = 0; i < ETISS_MAX_RESOURCES; i++)
    {
        if (cpu_->resources[i])
        {
            std::cout << "Resource Usage " << cpu_->resources[i] << ": " << cpu_->resourceUsages[i] << " cycles, "
                      << ((cpu_->resourceUsages[i] / (double)max) * 100) << "%" << std::endl;
        }
    }
#if ETISS_CPUCORE_DBG_APPROXIMATE_INSTRUCTION_COUNTER
    etiss::log(etiss::INFO, std::string("InstructionCounter: ") +
                                etiss::toString(instrcounter / ((double)cpu_->cpuTime_ps / 1000000.0)));
    etiss::log(etiss::INFO, std::string("MIPS (good estimation): ") +
                                etiss::toString(instrcounter / ((double)cpu_->cpuTime_ps / 1000000.0)));
#endif
 
    // cleanup plugins
    for (auto &p : plugins)
    {
        if (p)
        {
            p->cleanup();
            p->plugin_cpu_ = nullptr;
            p->plugin_system_ = nullptr;
            p->plugin_arch_ = nullptr;
        }
    }
 
    // undo system wrapping
    for (auto &syswrapper : syswrappers)
    {
        auto psys = syswrapper->unwrap(cpu_, system);
        if (psys)
        {
            system = psys;
        }
        else
        {
            std::stringstream stream;
            stream << "SERVE WARNING: SystemWrapperPlugin \"" << syswrapper->getPluginName()
                   << "\" failed to unwrap ETISS_System instance. Most likely results in a memory leak.";
            etiss::log(etiss::WARNING, stream.str());
            break;
        }
    }
 
    if (listener)
    {
        vcpu_->foreachField(
            [listener](std::shared_ptr<etiss::VirtualStruct::Field> f) { f->removeListener(listener); });
 
        delete listener;
    }
 
    return exception;
}