changes the io_buf

this relates to https://github.com/Minres/DBT-RISE-RISCV/issues/8 :
Debugger loses control when trap vector fetch fails

and https://github.com/Minres/DBT-RISE-RISCV/issues/7 : Two debugger
single-steps are required at reset vector

CMakeLists.txt

@@ -20,6 +20,7 @@ set(LIB_SOURCES
     src/iss/arch/tgc5c.cpp
     src/vm/interp/vm_tgc5c.cpp
     src/vm/fp_functions.cpp
+    src/iss/debugger/csr_names.cpp
     src/iss/semihosting/semihosting.cpp
 )
 
@@ -108,16 +109,6 @@ if(TARGET yaml-cpp::yaml-cpp)
     target_link_libraries(${PROJECT_NAME} PUBLIC yaml-cpp::yaml-cpp)
 endif()
 
-if(WITH_LLVM)
-    find_package(LLVM)
-    target_compile_definitions(${PROJECT_NAME} PUBLIC ${LLVM_DEFINITIONS})
-    target_include_directories(${PROJECT_NAME} PUBLIC ${LLVM_INCLUDE_DIRS})
-
-    if(BUILD_SHARED_LIBS)
-        target_link_libraries(${PROJECT_NAME} PUBLIC ${LLVM_LIBRARIES})
-    endif()
-endif()
-
 set_target_properties(${PROJECT_NAME} PROPERTIES
     VERSION ${PROJECT_VERSION}
     FRAMEWORK FALSE
@@ -261,3 +252,9 @@ if(TARGET scc-sysc)
         INCLUDES DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} # headers
     )
 endif()
+
+project(elfio-test)
+find_package(Boost COMPONENTS program_options thread REQUIRED)
+
+add_executable(${PROJECT_NAME} src/elfio.cpp)
+target_link_libraries(${PROJECT_NAME} PUBLIC elfio::elfio)

gen_input/templates/CORENAME.h.gtl

@@ -131,8 +131,6 @@ struct ${coreDef.name.toLowerCase()}: public arch_if {
 
     uint8_t* get_regs_base_ptr() override;
 
-    inline uint64_t get_icount() { return reg.icount; }
-
     inline bool should_stop() { return interrupt_sim; }
 
     inline uint64_t stop_code() { return interrupt_sim; }
@@ -141,8 +139,6 @@ struct ${coreDef.name.toLowerCase()}: public arch_if {
 
     virtual iss::sync_type needed_sync() const { return iss::NO_SYNC; }
 
-    inline uint32_t get_last_branch() { return reg.last_branch; }
-
 
 #pragma pack(push, 1)
     struct ${coreDef.name}_regs {<%

gen_input/templates/CORENAME_sysc.cpp.gtl

@@ -45,17 +45,17 @@ namespace interp {
 using namespace sysc;
 volatile std::array<bool, ${array_count}> ${coreDef.name.toLowerCase()}_init = {
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|m_p|interp", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_m_p<arch::${coreDef.name.toLowerCase()}>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         }),
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|mu_p|interp", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::${coreDef.name.toLowerCase()}>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         })<%if(coreDef.name.toLowerCase()=="tgc5d" || coreDef.name.toLowerCase()=="tgc5e") {%>,
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|mu_p_clic_pmp|interp", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::${coreDef.name.toLowerCase()}, (iss::arch::features_e)(iss::arch::FEAT_PMP | iss::arch::FEAT_EXT_N | iss::arch::FEAT_CLIC)>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         })<%}%>
@@ -66,17 +66,17 @@ namespace llvm {
 using namespace sysc;
 volatile std::array<bool, ${array_count}> ${coreDef.name.toLowerCase()}_init = {
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|m_p|llvm", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_m_p<arch::${coreDef.name.toLowerCase()}>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         }),
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|mu_p|llvm", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::${coreDef.name.toLowerCase()}>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         })<%if(coreDef.name.toLowerCase()=="tgc5d" || coreDef.name.toLowerCase()=="tgc5e") {%>,
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|mu_p_clic_pmp|llvm", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::${coreDef.name.toLowerCase()}, (iss::arch::features_e)(iss::arch::FEAT_PMP | iss::arch::FEAT_EXT_N | iss::arch::FEAT_CLIC)>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         })<%}%>
@@ -88,17 +88,17 @@ namespace tcc {
 using namespace sysc;
 volatile std::array<bool, ${array_count}> ${coreDef.name.toLowerCase()}_init = {
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|m_p|tcc", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_m_p<arch::${coreDef.name.toLowerCase()}>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         }),
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|mu_p|tcc", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::${coreDef.name.toLowerCase()}>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         })<%if(coreDef.name.toLowerCase()=="tgc5d" || coreDef.name.toLowerCase()=="tgc5e") {%>,
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|mu_p_clic_pmp|tcc", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::${coreDef.name.toLowerCase()}, (iss::arch::features_e)(iss::arch::FEAT_PMP | iss::arch::FEAT_EXT_N | iss::arch::FEAT_CLIC)>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         })<%}%>
@@ -110,17 +110,17 @@ namespace asmjit {
 using namespace sysc;
 volatile std::array<bool, ${array_count}> ${coreDef.name.toLowerCase()}_init = {
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|m_p|asmjit", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_m_p<arch::${coreDef.name.toLowerCase()}>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         }),
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|mu_p|asmjit", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::${coreDef.name.toLowerCase()}>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         })<%if(coreDef.name.toLowerCase()=="tgc5d" || coreDef.name.toLowerCase()=="tgc5e") {%>,
         iss_factory::instance().register_creator("${coreDef.name.toLowerCase()}|mu_p_clic_pmp|asmjit", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-            auto* cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+            auto* cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
             auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::${coreDef.name.toLowerCase()}, (iss::arch::features_e)(iss::arch::FEAT_PMP | iss::arch::FEAT_EXT_N | iss::arch::FEAT_CLIC)>>(cc);
             return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::${coreDef.name.toLowerCase()}*>(cpu), gdb_port)}};
         })<%}%>

gen_input/templates/asmjit/CORENAME.cpp.gtl

@@ -38,7 +38,9 @@
 #include <asmjit/asmjit.h>
 #include <util/logging.h>
 #include <iss/instruction_decoder.h>
-
+<%def fcsr = registers.find {it.name=='FCSR'}
+if(fcsr != null) {%>
+#include <vm/fp_functions.h><%}%>
 #ifndef FMT_HEADER_ONLY
 #define FMT_HEADER_ONLY
 #endif
@@ -88,23 +90,24 @@ protected:
     using super::write_reg_to_mem;
     using super::gen_read_mem;
     using super::gen_write_mem;
-    using super::gen_wait;
     using super::gen_leave;
     using super::gen_sync;
    
     using this_class = vm_impl<ARCH>;
     using compile_func = continuation_e (this_class::*)(virt_addr_t&, code_word_t, jit_holder&);
 
-    continuation_e gen_single_inst_behavior(virt_addr_t&, unsigned int &, jit_holder&) override;
+    continuation_e gen_single_inst_behavior(virt_addr_t&, jit_holder&) override;
     enum globals_e {TVAL = 0, GLOBALS_SIZE};
     void gen_block_prologue(jit_holder& jh) override;
     void gen_block_epilogue(jit_holder& jh) override;
     inline const char *name(size_t index){return traits::reg_aliases.at(index);}
-
+<%if(fcsr != null) {%>
+    inline const char *fname(size_t index){return index < 32?name(index+traits::F0):"illegal";}   
+<%}%>
     void gen_instr_prologue(jit_holder& jh);
     void gen_instr_epilogue(jit_holder& jh);
     inline void gen_raise(jit_holder& jh, uint16_t trap_id, uint16_t cause);
-    template <typename T, typename = std::enable_if_t<std::is_integral_v<T>>> void gen_set_tval(jit_holder& jh, T new_tval) ;
+    template <typename T, typename = typename std::enable_if<std::is_integral<T>::value>::type> void gen_set_tval(jit_holder& jh, T new_tval) ;
     void gen_set_tval(jit_holder& jh, x86_reg_t _new_tval) ;
 
     template<unsigned W, typename U, typename S = typename std::make_signed<U>::type>
@@ -112,7 +115,10 @@ protected:
         auto mask = (1ULL<<W) - 1;
         auto sign_mask = 1ULL<<(W-1);
         return (from & mask) | ((from & sign_mask) ? ~mask : 0);
-    } 
+    }
+<%functions.each{ it.eachLine { %>
+    ${it}<%}%>
+<%}%>
 private:
     /****************************************************************************
      * start opcode definitions
@@ -195,7 +201,7 @@ private:
         gen_raise(jh, 0, 2);
         gen_sync(jh, POST_SYNC, instr_descr.size());
         gen_instr_epilogue(jh);
-        return BRANCH;
+        return ILLEGAL_INSTR;
     }
 };
 
@@ -215,7 +221,7 @@ vm_impl<ARCH>::vm_impl(ARCH &core, unsigned core_id, unsigned cluster_id)
     }()) {}
 
 template <typename ARCH>
-continuation_e vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, unsigned int &inst_cnt, jit_holder& jh) {
+continuation_e vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, jit_holder& jh) {
     enum {TRAP_ID=1<<16};
     code_word_t instr = 0;
     phys_addr_t paddr(pc);
@@ -224,10 +230,9 @@ continuation_e vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, unsigned
         paddr = this->core.virt2phys(pc);
     auto res = this->core.read(paddr, 4, data);
     if (res != iss::Ok)
-        throw trap_access(TRAP_ID, pc.val);
+        return ILLEGAL_FETCH;
     if (instr == 0x0000006f || (instr&0xffff)==0xa001)
-        throw simulation_stopped(0); // 'J 0' or 'C.J 0'
-    ++inst_cnt;
+        return JUMP_TO_SELF;
     uint32_t inst_index = instr_decoder.decode_instr(instr);
     compile_func f = nullptr;
     if(inst_index < instr_descr.size())
@@ -257,6 +262,7 @@ void vm_impl<ARCH>::gen_instr_epilogue(jit_holder& jh) {
     cmp(cc, current_trap_state, 0);
     cc.jne(jh.trap_entry);
     cc.inc(get_ptr_for(jh, traits::ICOUNT));
+    cc.inc(get_ptr_for(jh, traits::CYCLE));
 }
 template <typename ARCH>
 void vm_impl<ARCH>::gen_block_prologue(jit_holder& jh){
@@ -302,6 +308,7 @@ inline void vm_impl<ARCH>::gen_raise(jit_holder& jh, uint16_t trap_id, uint16_t
     auto tmp1 = get_reg_for(cc, traits::TRAP_STATE);
     mov(cc, tmp1, 0x80ULL << 24 | (cause << 16) | trap_id);
     mov(cc, get_ptr_for(jh, traits::TRAP_STATE), tmp1);
+    cc.jmp(jh.trap_entry);
 }
 template <typename ARCH>
 template <typename T, typename>
@@ -310,8 +317,8 @@ void vm_impl<ARCH>::gen_set_tval(jit_holder& jh, T new_tval) {
     }
 template <typename ARCH>
 void vm_impl<ARCH>::gen_set_tval(jit_holder& jh, x86_reg_t _new_tval) {
-    if(std::holds_alternative<x86::Gp>(_new_tval)) {
-        x86::Gp new_tval = std::get<x86::Gp>(_new_tval);
+    if(nonstd::holds_alternative<x86::Gp>(_new_tval)) {
+        x86::Gp new_tval = nonstd::get<x86::Gp>(_new_tval);
         if(new_tval.size() < 8)
             new_tval = gen_ext_Gp(jh.cc, new_tval, 64, false);
         mov(jh.cc, jh.globals[TVAL], new_tval);

gen_input/templates/interp/CORENAME.cpp.gtl

@@ -199,9 +199,6 @@ template <typename CODE_WORD> void debug_fn(CODE_WORD insn) {
     volatile CODE_WORD x = insn;
     insn = 2 * x;
 }
-
-template <typename ARCH> vm_impl<ARCH>::vm_impl() { this(new ARCH()); }
-
 // according to
 // https://stackoverflow.com/questions/8871204/count-number-of-1s-in-binary-representation
 #ifdef __GCC__
@@ -257,17 +254,21 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
     while(!this->core.should_stop() &&
             !(is_icount_limit_enabled(cond) && icount >= count_limit) &&
             !(is_fcount_limit_enabled(cond) && fetch_count >= count_limit)){
-        fetch_count++;
+        if(this->debugging_enabled())
+            this->tgt_adapter->check_continue(*PC);
+        pc.val=*PC;
         if(fetch_ins(pc, data)!=iss::Ok){
-            this->do_sync(POST_SYNC, std::numeric_limits<unsigned>::max());
-            pc.val = super::core.enter_trap(std::numeric_limits<uint64_t>::max(), pc.val, 0);
+            if(this->sync_exec && PRE_SYNC) this->do_sync(PRE_SYNC, std::numeric_limits<unsigned>::max());
+            process_spawn_blocks();
+            if(this->sync_exec && POST_SYNC) this->do_sync(PRE_SYNC, std::numeric_limits<unsigned>::max());
+            pc.val = super::core.enter_trap(arch::traits<ARCH>::RV_CAUSE_FETCH_ACCESS<<16, pc.val, 0);
         } else {
             if (is_jump_to_self_enabled(cond) &&
                     (instr == 0x0000006f || (instr&0xffff)==0xa001)) throw simulation_stopped(0); // 'J 0' or 'C.J 0'
             uint32_t inst_index = instr_decoder.decode_instr(instr);
             opcode_e inst_id = arch::traits<ARCH>::opcode_e::MAX_OPCODE;;
             if(inst_index <instr_descr.size())
-                inst_id = instr_descr.at(instr_decoder.decode_instr(instr)).op;
+                inst_id = instr_descr[inst_index].op;
 
             // pre execution stuff
             this->core.reg.last_branch = 0;
@@ -279,6 +280,7 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                     <%}%>if(this->disass_enabled){
                         /* generate console output when executing the command */<%instr.disass.eachLine{%>
                         ${it}<%}%>
+                        this->core.disass_output(pc.val, mnemonic);
                     }
                     // used registers<%instr.usedVariables.each{ k,v->
                     if(v.isArray) {%>
@@ -310,11 +312,11 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                 icount++;
                 instret++;
             }
-            cycle++;
-            pc.val=*NEXT_PC;
-            this->core.reg.PC = this->core.reg.NEXT_PC;
+            *PC = *NEXT_PC;
             this->core.reg.trap_state =  this->core.reg.pending_trap;
         }
+        fetch_count++;
+        cycle++;
     }
     return pc;
 }

gen_input/templates/llvm/CORENAME.cpp.gtl

@@ -37,7 +37,9 @@
 #include <iss/llvm/vm_base.h>
 #include <util/logging.h>
 #include <iss/instruction_decoder.h>
-
+<%def fcsr = registers.find {it.name=='FCSR'}
+if(fcsr != null) {%>
+#include <vm/fp_functions.h><%}%>
 #ifndef FMT_HEADER_ONLY
 #define FMT_HEADER_ONLY
 #endif
@@ -83,7 +85,9 @@ protected:
     using vm_base<ARCH>::get_reg_ptr;
 
     inline const char *name(size_t index){return traits::reg_aliases.at(index);}
-
+<%if(fcsr != null) {%>
+    inline const char *fname(size_t index){return index < 32?name(index+traits::F0):"illegal";}   
+<%}%>
     template <typename T> inline ConstantInt *size(T type) {
         return ConstantInt::get(getContext(), APInt(32, type->getType()->getScalarSizeInBits()));
     }
@@ -97,7 +101,7 @@ protected:
         return super::gen_cond_assign(cond, this->gen_ext(trueVal, size), this->gen_ext(falseVal, size));
     }
 
-    std::tuple<continuation_e, BasicBlock *> gen_single_inst_behavior(virt_addr_t &, unsigned int &, BasicBlock *) override;
+    std::tuple<continuation_e, BasicBlock *> gen_single_inst_behavior(virt_addr_t &, BasicBlock *) override;
 
     void gen_leave_behavior(BasicBlock *leave_blk) override;
     void gen_raise_trap(uint16_t trap_id, uint16_t cause);
@@ -130,8 +134,10 @@ protected:
         auto mask = (1ULL<<W) - 1;
         auto sign_mask = 1ULL<<(W-1);
         return (from & mask) | ((from & sign_mask) ? ~mask : 0);
-    }   
-
+    }
+<%functions.each{ it.eachLine { %>
+    ${it}<%}%>
+<%}%>
 private:
     /****************************************************************************
      * start opcode definitions
@@ -198,7 +204,7 @@ private:
             };
             this->builder.CreateCall(this->mod->getFunction("print_disass"), args);
         }
-		this->gen_sync(iss::PRE_SYNC, instr_descr.size());
+        this->gen_sync(iss::PRE_SYNC, instr_descr.size());
         this->builder.CreateStore(this->builder.CreateLoad(this->get_typeptr(traits::NEXT_PC), get_reg_ptr(traits::NEXT_PC), true),
                                    get_reg_ptr(traits::PC), true);
         this->builder.CreateStore(
@@ -212,7 +218,7 @@ private:
         bb = this->leave_blk;
         this->gen_instr_epilogue(bb);
         this->builder.CreateBr(bb);
-        return std::make_tuple(BRANCH, nullptr);
+        return std::make_tuple(ILLEGAL_INSTR, nullptr);
     }    
 };
 
@@ -238,7 +244,7 @@ vm_impl<ARCH>::vm_impl(ARCH &core, unsigned core_id, unsigned cluster_id)
 
 template <typename ARCH>
 std::tuple<continuation_e, BasicBlock *>
-vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, unsigned int &inst_cnt, BasicBlock *this_block) {
+vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, BasicBlock *this_block) {
     // we fetch at max 4 byte, alignment is 2
     enum {TRAP_ID=1<<16};
     code_word_t instr = 0;
@@ -247,20 +253,13 @@ vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, unsigned int &inst_cnt,
     auto *const data = (uint8_t *)&instr;
     if(this->core.has_mmu())
         paddr = this->core.virt2phys(pc);
-    //TODO: re-add page handling
-//    if ((pc.val & upper_bits) != ((pc.val + 2) & upper_bits)) { // we may cross a page boundary
-//        auto res = this->core.read(paddr, 2, data);
-//        if (res != iss::Ok) throw trap_access(TRAP_ID, pc.val);
-//        if ((instr & 0x3) == 0x3) { // this is a 32bit instruction
-//            res = this->core.read(this->core.v2p(pc + 2), 2, data + 2);
-//        }
-//    } else {
     auto res = this->core.read(paddr, 4, data);
-    if (res != iss::Ok) throw trap_access(TRAP_ID, pc.val);
-//    }
-    if (instr == 0x0000006f || (instr&0xffff)==0xa001) throw simulation_stopped(0); // 'J 0' or 'C.J 0'
-    // curr pc on stack
-    ++inst_cnt;
+    if (res != iss::Ok) 
+        return std::make_tuple(ILLEGAL_FETCH, nullptr);
+    if (instr == 0x0000006f || (instr&0xffff)==0xa001){
+        this->builder.CreateBr(this->leave_blk);
+        return std::make_tuple(JUMP_TO_SELF, nullptr);
+        }
     uint32_t inst_index = instr_decoder.decode_instr(instr);
     compile_func f = nullptr;
     if(inst_index < instr_descr.size())
@@ -281,6 +280,7 @@ template <typename ARCH>
 void vm_impl<ARCH>::gen_raise_trap(uint16_t trap_id, uint16_t cause) {
     auto *TRAP_val = this->gen_const(32, 0x80 << 24 | (cause << 16) | trap_id);
     this->builder.CreateStore(TRAP_val, get_reg_ptr(traits::TRAP_STATE), true);
+    this->builder.CreateBr(this->trap_blk);
 }
 
 template <typename ARCH>
@@ -341,6 +341,10 @@ void vm_impl<ARCH>::gen_instr_epilogue(BasicBlock *bb) {
     auto* icount_val = this->builder.CreateAdd(
         this->builder.CreateLoad(this->get_typeptr(arch::traits<ARCH>::ICOUNT), get_reg_ptr(arch::traits<ARCH>::ICOUNT)), this->gen_const(64U, 1));
     this->builder.CreateStore(icount_val, get_reg_ptr(arch::traits<ARCH>::ICOUNT), false);
+    //increment cyclecount
+    auto* cycle_val = this->builder.CreateAdd(
+        this->builder.CreateLoad(this->get_typeptr(arch::traits<ARCH>::CYCLE), get_reg_ptr(arch::traits<ARCH>::CYCLE)), this->gen_const(64U, 1));
+    this->builder.CreateStore(cycle_val, get_reg_ptr(arch::traits<ARCH>::CYCLE), false);
 }
 
 } // namespace ${coreDef.name.toLowerCase()}
@@ -383,4 +387,4 @@ volatile std::array<bool, 2> dummy = {
 };
 }
 }
-// clang-format on
+// clang-format on

gen_input/templates/tcc/CORENAME.cpp.gtl

@@ -38,7 +38,9 @@
 #include <util/logging.h>
 #include <sstream>
 #include <iss/instruction_decoder.h>
-
+<%def fcsr = registers.find {it.name=='FCSR'}
+if(fcsr != null) {%>
+#include <vm/fp_functions.h><%}%>
 #ifndef FMT_HEADER_ONLY
 #define FMT_HEADER_ONLY
 #endif
@@ -81,16 +83,21 @@ protected:
     using vm_base<ARCH>::get_reg_ptr;
 
     using this_class = vm_impl<ARCH>;
-    using compile_ret_t = std::tuple<continuation_e>;
+    using compile_ret_t = continuation_e;
     using compile_func = compile_ret_t (this_class::*)(virt_addr_t &pc, code_word_t instr, tu_builder&);
 
     inline const char *name(size_t index){return traits::reg_aliases.at(index);}
+<%
+if(fcsr != null) {%>
+    inline const char *fname(size_t index){return index < 32?name(index+traits::F0):"illegal";}   
+<%}%>
+    void add_prologue(tu_builder& tu) override;
 
     void setup_module(std::string m) override {
         super::setup_module(m);
     }
 
-    compile_ret_t gen_single_inst_behavior(virt_addr_t &, unsigned int &, tu_builder&) override;
+    compile_ret_t gen_single_inst_behavior(virt_addr_t &, tu_builder&) override;
 
     void gen_trap_behavior(tu_builder& tu) override;
 
@@ -98,8 +105,6 @@ protected:
 
     void gen_leave_trap(tu_builder& tu, unsigned lvl);
 
-    void gen_wait(tu_builder& tu, unsigned type);
-
     inline void gen_set_tval(tu_builder& tu, uint64_t new_tval);
 
     inline void gen_set_tval(tu_builder& tu, value new_tval);
@@ -133,6 +138,9 @@ protected:
         return (from & mask) | ((from & sign_mask) ? ~mask : 0);
     }
 
+<%functions.each{ it.eachLine { %>
+    ${it}<%}%>
+<%}%>
 private:
     /****************************************************************************
      * start opcode definitions
@@ -163,10 +171,12 @@ private:
         <%}%>if(this->disass_enabled){
             /* generate console output when executing the command */<%instr.disass.eachLine{%>
             ${it}<%}%>
+            tu("print_disass(core_ptr, {:#x}, \"{}\");", pc.val, mnemonic);
         }
         auto cur_pc_val = tu.constant(pc.val, traits::reg_bit_widths[traits::PC]);
         pc=pc+ ${instr.length/8};
         gen_set_pc(tu, pc, traits::NEXT_PC);
+        tu("(*cycle)++;");
         tu.open_scope();
         this->gen_set_tval(tu, instr);
         <%instr.behavior.eachLine{%>${it}
@@ -187,11 +197,11 @@ private:
             tu("print_disass(core_ptr, {:#x}, \"{}\");", pc.val, std::string("illegal_instruction"));
         }
         pc = pc + ((instr & 3) == 3 ? 4 : 2);
-        gen_raise_trap(tu, 0, 2);     // illegal instruction trap
+        gen_raise_trap(tu, 0, static_cast<int32_t>(traits:: RV_CAUSE_ILLEGAL_INSTRUCTION));
         this->gen_set_tval(tu, instr);
         vm_impl::gen_sync(tu, iss::POST_SYNC, instr_descr.size());
         vm_impl::gen_trap_check(tu);
-        return BRANCH;
+        return ILLEGAL_INSTR;
     }
 };
 
@@ -216,28 +226,19 @@ vm_impl<ARCH>::vm_impl(ARCH &core, unsigned core_id, unsigned cluster_id)
     }()) {}
 
 template <typename ARCH>
-std::tuple<continuation_e>
-vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, unsigned int &inst_cnt, tu_builder& tu) {
+continuation_e
+vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, tu_builder& tu) {
     // we fetch at max 4 byte, alignment is 2
     enum {TRAP_ID=1<<16};
     code_word_t instr = 0;
     phys_addr_t paddr(pc);
     if(this->core.has_mmu())
         paddr = this->core.virt2phys(pc);
-    //TODO: re-add page handling
-//    if ((pc.val & upper_bits) != ((pc.val + 2) & upper_bits)) { // we may cross a page boundary
-//        auto res = this->core.read(paddr, 2, data);
-//        if (res != iss::Ok) throw trap_access(TRAP_ID, pc.val);
-//        if ((insn & 0x3) == 0x3) { // this is a 32bit instruction
-//            res = this->core.read(this->core.v2p(pc + 2), 2, data + 2);
-//        }
-//    } else {
     auto res = this->core.read(paddr, 4, reinterpret_cast<uint8_t*>(&instr));
-    if (res != iss::Ok) throw trap_access(TRAP_ID, pc.val);
-//    }
-    if (instr == 0x0000006f || (instr&0xffff)==0xa001) throw simulation_stopped(0); // 'J 0' or 'C.J 0'
-    // curr pc on stack
-    ++inst_cnt;
+    if (res != iss::Ok)
+        return ILLEGAL_FETCH;
+    if (instr == 0x0000006f || (instr&0xffff)==0xa001) 
+        return JUMP_TO_SELF;
     uint32_t inst_index = instr_decoder.decode_instr(instr);
     compile_func f = nullptr;
     if(inst_index < instr_descr.size())
@@ -258,9 +259,6 @@ template <typename ARCH> void vm_impl<ARCH>::gen_leave_trap(tu_builder& tu, unsi
     tu.store(traits::LAST_BRANCH, tu.constant(static_cast<int>(UNKNOWN_JUMP), 32));
 }
 
-template <typename ARCH> void vm_impl<ARCH>::gen_wait(tu_builder& tu, unsigned type) {
-}
-
 template <typename ARCH> void vm_impl<ARCH>::gen_set_tval(tu_builder& tu, uint64_t new_tval) {
     tu(fmt::format("tval = {};", new_tval));
 }
@@ -275,6 +273,41 @@ template <typename ARCH> void vm_impl<ARCH>::gen_trap_behavior(tu_builder& tu) {
     tu.store(traits::LAST_BRANCH, tu.constant(static_cast<int>(UNKNOWN_JUMP),32));
     tu("return *next_pc;");
 }
+template <typename ARCH> void vm_impl<ARCH>::add_prologue(tu_builder& tu){
+    std::ostringstream os;
+    os << tu.add_reg_ptr("trap_state", arch::traits<ARCH>::TRAP_STATE, this->regs_base_ptr);
+    os << tu.add_reg_ptr("pending_trap", arch::traits<ARCH>::PENDING_TRAP, this->regs_base_ptr);
+    os << tu.add_reg_ptr("cycle", arch::traits<ARCH>::CYCLE, this->regs_base_ptr);
+<%if(fcsr != null) {%>
+    os << "uint32_t (*fget_flags)()=" << (uintptr_t)&fget_flags << ";\\n";
+    os << "uint32_t (*fadd_s)(uint32_t v1, uint32_t v2, uint8_t mode)=" << (uintptr_t)&fadd_s << ";\\n";
+    os << "uint32_t (*fsub_s)(uint32_t v1, uint32_t v2, uint8_t mode)=" << (uintptr_t)&fsub_s << ";\\n";
+    os << "uint32_t (*fmul_s)(uint32_t v1, uint32_t v2, uint8_t mode)=" << (uintptr_t)&fmul_s << ";\\n";
+    os << "uint32_t (*fdiv_s)(uint32_t v1, uint32_t v2, uint8_t mode)=" << (uintptr_t)&fdiv_s << ";\\n";
+    os << "uint32_t (*fsqrt_s)(uint32_t v1, uint8_t mode)=" << (uintptr_t)&fsqrt_s << ";\\n";
+    os << "uint32_t (*fcmp_s)(uint32_t v1, uint32_t v2, uint32_t op)=" << (uintptr_t)&fcmp_s << ";\\n";
+    os << "uint32_t (*fcvt_s)(uint32_t v1, uint32_t op, uint8_t mode)=" << (uintptr_t)&fcvt_s << ";\\n";
+    os << "uint32_t (*fmadd_s)(uint32_t v1, uint32_t v2, uint32_t v3, uint32_t op, uint8_t mode)=" << (uintptr_t)&fmadd_s << ";\\n";
+    os << "uint32_t (*fsel_s)(uint32_t v1, uint32_t v2, uint32_t op)=" << (uintptr_t)&fsel_s << ";\\n";
+    os << "uint32_t (*fclass_s)( uint32_t v1 )=" << (uintptr_t)&fclass_s << ";\\n";
+    os << "uint32_t (*fconv_d2f)(uint64_t v1, uint8_t mode)=" << (uintptr_t)&fconv_d2f << ";\\n";
+    os << "uint64_t (*fconv_f2d)(uint32_t v1, uint8_t mode)=" << (uintptr_t)&fconv_f2d << ";\\n";
+    os << "uint64_t (*fadd_d)(uint64_t v1, uint64_t v2, uint8_t mode)=" << (uintptr_t)&fadd_d << ";\\n";
+    os << "uint64_t (*fsub_d)(uint64_t v1, uint64_t v2, uint8_t mode)=" << (uintptr_t)&fsub_d << ";\\n";
+    os << "uint64_t (*fmul_d)(uint64_t v1, uint64_t v2, uint8_t mode)=" << (uintptr_t)&fmul_d << ";\\n";
+    os << "uint64_t (*fdiv_d)(uint64_t v1, uint64_t v2, uint8_t mode)=" << (uintptr_t)&fdiv_d << ";\\n";
+    os << "uint64_t (*fsqrt_d)(uint64_t v1, uint8_t mode)=" << (uintptr_t)&fsqrt_d << ";\\n";
+    os << "uint64_t (*fcmp_d)(uint64_t v1, uint64_t v2, uint32_t op)=" << (uintptr_t)&fcmp_d << ";\\n";
+    os << "uint64_t (*fcvt_d)(uint64_t v1, uint32_t op, uint8_t mode)=" << (uintptr_t)&fcvt_d << ";\\n";
+    os << "uint64_t (*fmadd_d)(uint64_t v1, uint64_t v2, uint64_t v3, uint32_t op, uint8_t mode)=" << (uintptr_t)&fmadd_d << ";\\n";
+    os << "uint64_t (*fsel_d)(uint64_t v1, uint64_t v2, uint32_t op)=" << (uintptr_t)&fsel_d << ";\\n";
+    os << "uint64_t (*fclass_d)(uint64_t v1  )=" << (uintptr_t)&fclass_d << ";\\n";
+    os << "uint64_t (*fcvt_32_64)(uint32_t v1, uint32_t op, uint8_t mode)=" << (uintptr_t)&fcvt_32_64 << ";\\n";
+    os << "uint32_t (*fcvt_64_32)(uint64_t v1, uint32_t op, uint8_t mode)=" << (uintptr_t)&fcvt_64_32 << ";\\n";
+    os << "uint32_t (*unbox_s)(uint64_t v)=" << (uintptr_t)&unbox_s << ";\\n";
+    <%}%>
+    tu.add_prologue(os.str());
+}
 
 } // namespace ${coreDef.name.toLowerCase()}
 

src/elfio.cpp

@@ -0,0 +1,35 @@
+#ifdef _MSC_VER
+#define _SCL_SECURE_NO_WARNINGS
+#define ELFIO_NO_INTTYPES
+#endif
+
+#include <elfio/elfio_dump.hpp>
+#include <iostream>
+
+using namespace ELFIO;
+
+int main(int argc, char** argv) {
+    if(argc != 2) {
+        printf("Usage: elfdump <file_name>\n");
+        return 1;
+    }
+
+    elfio reader;
+
+    if(!reader.load(argv[1])) {
+        printf("File %s is not found or it is not an ELF file\n", argv[1]);
+        return 1;
+    }
+
+    dump::header(std::cout, reader);
+    dump::section_headers(std::cout, reader);
+    dump::segment_headers(std::cout, reader);
+    dump::symbol_tables(std::cout, reader);
+    dump::notes(std::cout, reader);
+    dump::modinfo(std::cout, reader);
+    dump::dynamic_tags(std::cout, reader);
+    dump::section_datas(std::cout, reader);
+    dump::segment_datas(std::cout, reader);
+
+    return 0;
+}

src/iss/arch/hwl.h

@@ -51,8 +51,8 @@ public:
     virtual ~hwl() = default;
 
 protected:
-    iss::status read_custom_csr_reg(unsigned addr, reg_t& val) override;
-    iss::status write_custom_csr_reg(unsigned addr, reg_t val) override;
+    iss::status read_custom_csr(unsigned addr, reg_t& val) override;
+    iss::status write_custom_csr(unsigned addr, reg_t val) override;
 };
 
 template <typename BASE>
@@ -68,7 +68,7 @@ inline hwl<BASE>::hwl(feature_config cfg)
     }
 }
 
-template <typename BASE> inline iss::status iss::arch::hwl<BASE>::read_custom_csr_reg(unsigned addr, reg_t& val) {
+template <typename BASE> inline iss::status iss::arch::hwl<BASE>::read_custom_csr(unsigned addr, reg_t& val) {
     switch(addr) {
     case 0x800:
         val = this->reg.lpstart0;
@@ -92,7 +92,7 @@ template <typename BASE> inline iss::status iss::arch::hwl<BASE>::read_custom_cs
     return iss::Ok;
 }
 
-template <typename BASE> inline iss::status iss::arch::hwl<BASE>::write_custom_csr_reg(unsigned addr, reg_t val) {
+template <typename BASE> inline iss::status iss::arch::hwl<BASE>::write_custom_csr(unsigned addr, reg_t val) {
     switch(addr) {
     case 0x800:
         this->reg.lpstart0 = val;

src/iss/arch/riscv_hart_common.h

@@ -35,11 +35,15 @@
 #ifndef _RISCV_HART_COMMON
 #define _RISCV_HART_COMMON
 
+#include "iss/vm_types.h"
+#include <array>
 #include <cstdint>
 #include <elfio/elfio.hpp>
 #include <fmt/format.h>
 #include <iss/arch_if.h>
 #include <iss/log_categories.h>
+#include <limits>
+#include <sstream>
 #include <string>
 #include <unordered_map>
 #include <util/logging.h>
@@ -55,8 +59,6 @@
 namespace iss {
 namespace arch {
 
-enum { tohost_dflt = 0xF0001000, fromhost_dflt = 0xF0001040 };
-
 enum features_e { FEAT_NONE, FEAT_PMP = 1, FEAT_EXT_N = 2, FEAT_CLIC = 4, FEAT_DEBUG = 8, FEAT_TCM = 16 };
 
 enum riscv_csr {
@@ -312,58 +314,101 @@ inline void write_reg_uint32(uint64_t offs, uint32_t& reg, const uint8_t* const
 }
 struct riscv_hart_common {
     riscv_hart_common(){};
-    ~riscv_hart_common(){};
+    ~riscv_hart_common() {
+        if(io_buf.str().length()) {
+            CPPLOG(INFO) << "tohost send '" << io_buf.str() << "'";
+        }
+    };
     std::unordered_map<std::string, uint64_t> symbol_table;
+    uint64_t entry_address{0};
+    uint64_t tohost = std::numeric_limits<uint64_t>::max();
+    uint64_t fromhost = std::numeric_limits<uint64_t>::max();
+    std::stringstream io_buf;
 
-    std::unordered_map<std::string, uint64_t> get_sym_table(std::string name) {
-        if(!symbol_table.empty())
-            return symbol_table;
-        FILE* fp = fopen(name.c_str(), "r");
-        if(fp) {
-            std::array<char, 5> buf;
-            auto n = fread(buf.data(), 1, 4, fp);
-            fclose(fp);
-            if(n != 4)
-                throw std::runtime_error("input file has insufficient size");
-            buf[4] = 0;
-            if(strcmp(buf.data() + 1, "ELF") == 0) {
-                // Create elfio reader
-                ELFIO::elfio reader;
-                // Load ELF data
-                if(!reader.load(name))
-                    throw std::runtime_error("could not process elf file");
-                // check elf properties
-                if(reader.get_type() != ET_EXEC)
-                    throw std::runtime_error("wrong elf type in file");
-                if(reader.get_machine() != EM_RISCV)
-                    throw std::runtime_error("wrong elf machine in file");
-                const auto sym_sec = reader.sections[".symtab"];
-                if(SHT_SYMTAB == sym_sec->get_type() || SHT_DYNSYM == sym_sec->get_type()) {
-                    ELFIO::symbol_section_accessor symbols(reader, sym_sec);
-                    auto sym_no = symbols.get_symbols_num();
-                    std::string name;
-                    ELFIO::Elf64_Addr value = 0;
-                    ELFIO::Elf_Xword size = 0;
-                    unsigned char bind = 0;
-                    unsigned char type = 0;
-                    ELFIO::Elf_Half section = 0;
-                    unsigned char other = 0;
-                    for(auto i = 0U; i < sym_no; ++i) {
-                        symbols.get_symbol(i, name, value, size, bind, type, section, other);
-                        if(name != "") {
-                            this->symbol_table[name] = value;
+    bool read_elf_file(std::string name, uint8_t expected_elf_class,
+                       std::function<iss::status(uint64_t, uint64_t, const uint8_t* const)> cb) {
+        // Create elfio reader
+        ELFIO::elfio reader;
+        // Load ELF data
+        if(reader.load(name)) {
+            // check elf properties
+            if(reader.get_class() != expected_elf_class)
+                return false;
+            if(reader.get_type() != ELFIO::ET_EXEC)
+                return false;
+            if(reader.get_machine() != ELFIO::EM_RISCV)
+                return false;
+            entry_address = reader.get_entry();
+            for(const auto& pseg : reader.segments) {
+                const auto fsize = pseg->get_file_size(); // 0x42c/0x0
+                const auto seg_data = pseg->get_data();
+                const auto type = pseg->get_type();
+                if(type == 1 && fsize > 0) {
+                    auto res = cb(pseg->get_physical_address(), fsize, reinterpret_cast<const uint8_t* const>(seg_data));
+                    if(res != iss::Ok)
+                        CPPLOG(ERR) << "problem writing " << fsize << "bytes to 0x" << std::hex << pseg->get_physical_address();
+                }
+            }
+            const auto sym_sec = reader.sections[".symtab"];
+            if(ELFIO::SHT_SYMTAB == sym_sec->get_type() || ELFIO::SHT_DYNSYM == sym_sec->get_type()) {
+                ELFIO::symbol_section_accessor symbols(reader, sym_sec);
+                auto sym_no = symbols.get_symbols_num();
+                std::string name;
+                ELFIO::Elf64_Addr value = 0;
+                ELFIO::Elf_Xword size = 0;
+                unsigned char bind = 0;
+                unsigned char type = 0;
+                ELFIO::Elf_Half section = 0;
+                unsigned char other = 0;
+                for(auto i = 0U; i < sym_no; ++i) {
+                    symbols.get_symbol(i, name, value, size, bind, type, section, other);
+                    if(name != "") {
+                        this->symbol_table[name] = value;
 #ifndef NDEBUG
-                            CPPLOG(DEBUG) << "Found Symbol " << name;
+                        CPPLOG(DEBUG) << "Found Symbol " << name;
 #endif
-                        }
                     }
                 }
-                return symbol_table;
+                try {
+                    tohost = symbol_table.at("tohost");
+                } catch(std::out_of_range& e) {
+                }
+                try {
+                    fromhost = symbol_table.at("fromhost");
+                } catch(std::out_of_range& e) {
+                }
             }
-            throw std::runtime_error(fmt::format("memory load file {} is not a valid elf file", name));
-        } else
-            throw std::runtime_error(fmt::format("memory load file not found, check if {} is a valid file", name));
+            return true;
+        }
+        return false;
     };
+    iss::status execute_sys_write(arch_if* aif, const std::array<uint64_t, 8>& loaded_payload, unsigned mem_type) {
+        uint64_t fd = loaded_payload[1];
+        uint64_t buf_ptr = loaded_payload[2];
+        uint64_t len = loaded_payload[3];
+        std::vector<char> buf(len);
+        if(aif->read(address_type::PHYSICAL, access_type::DEBUG_READ, mem_type, buf_ptr, len, reinterpret_cast<uint8_t*>(buf.data()))) {
+            CPPLOG(ERR) << "SYS_WRITE buffer read went wrong";
+            return iss::Err;
+        }
+        // we disregard the fd and just log to stdout
+        for(size_t i = 0; i < len; i++) {
+            if(buf[i] == '\n' || buf[i] == '\0') {
+                CPPLOG(INFO) << "tohost send '" << io_buf.str() << "'";
+                io_buf.str("");
+            } else
+                io_buf << buf[i];
+        }
+
+        // Not sure what the correct return value should be
+        uint8_t ret_val = 1;
+        if(fromhost != std::numeric_limits<uint64_t>::max())
+            if(aif->write(address_type::PHYSICAL, access_type::DEBUG_WRITE, mem_type, fromhost, 1, &ret_val)) {
+                CPPLOG(ERR) << "Fromhost write went wrong";
+                return iss::Err;
+            }
+        return iss::Ok;
+    }
 };
 
 } // namespace arch

src/iss/arch/riscv_hart_m_p.h

@@ -41,6 +41,11 @@
 #include "iss/vm_if.h"
 #include "iss/vm_types.h"
 #include "riscv_hart_common.h"
+#include "util/logging.h"
+#include <algorithm>
+#include <cstdint>
+#include <elfio/elf_types.hpp>
+#include <limits>
 #include <stdexcept>
 #ifndef FMT_HEADER_ONLY
 #define FMT_HEADER_ONLY
@@ -278,7 +283,7 @@ public:
 
     void disass_output(uint64_t pc, const std::string instr) override {
         NSCLOG(INFO, LOGCAT) << fmt::format("0x{:016x}    {:40} [s:0x{:x};c:{}]", pc, instr, (reg_t)state.mstatus,
-                                            this->reg.icount + cycle_offset);
+                                            this->reg.cycle + cycle_offset);
     };
 
     iss::instrumentation_if* get_instrumentation_if() override { return &instr_if; }
@@ -311,7 +316,7 @@ protected:
 
         uint64_t get_pendig_traps() override { return arch.reg.trap_state; }
 
-        uint64_t get_total_cycles() override { return arch.reg.icount + arch.cycle_offset; }
+        uint64_t get_total_cycles() override { return arch.reg.cycle + arch.cycle_offset; }
 
         void update_last_instr_cycles(unsigned cycles) override { arch.cycle_offset += cycles - 1; }
 
@@ -321,7 +326,7 @@ protected:
 
         unsigned get_reg_size(unsigned num) override { return traits<BASE>::reg_bit_widths[num]; }
 
-        std::unordered_map<std::string, uint64_t> get_symbol_table(std::string name) override { return arch.get_sym_table(name); }
+        std::unordered_map<std::string, uint64_t> const& get_symbol_table(std::string name) override { return arch.symbol_table; }
 
         riscv_hart_m_p<BASE, FEAT, LOGCAT>& arch;
     };
@@ -343,9 +348,6 @@ protected:
     int64_t instret_offset{0};
     uint64_t minstret_csr{0};
     reg_t fault_data;
-    uint64_t tohost = tohost_dflt;
-    uint64_t fromhost = fromhost_dflt;
-    bool tohost_lower_written = false;
     riscv_instrumentation_if instr_if;
 
     semihosting_cb_t<reg_t> semihosting_cb;
@@ -355,7 +357,6 @@ protected:
     using csr_page_type = typename csr_type::page_type;
     mem_type mem;
     csr_type csr;
-    std::stringstream uart_buf;
     std::unordered_map<reg_t, uint64_t> ptw;
     std::unordered_map<uint64_t, uint8_t> atomic_reservation;
     std::unordered_map<unsigned, rd_csr_f> csr_rd_cb;
@@ -377,8 +378,8 @@ protected:
 
     std::vector<uint8_t> tcm;
 
-    iss::status read_csr_reg(unsigned addr, reg_t& val);
-    iss::status write_csr_reg(unsigned addr, reg_t val);
+    iss::status read_plain(unsigned addr, reg_t& val);
+    iss::status write_plain(unsigned addr, reg_t val);
     iss::status read_null(unsigned addr, reg_t& val);
     iss::status write_null(unsigned addr, reg_t val) { return iss::status::Ok; }
     iss::status read_cycle(unsigned addr, reg_t& val);
@@ -399,17 +400,19 @@ protected:
     iss::status read_intstatus(unsigned addr, reg_t& val);
     iss::status write_intthresh(unsigned addr, reg_t val);
     iss::status write_xtvt(unsigned addr, reg_t val);
-    iss::status write_dcsr_dcsr(unsigned addr, reg_t val);
-    iss::status read_dcsr_reg(unsigned addr, reg_t& val);
-    iss::status write_dcsr_reg(unsigned addr, reg_t val);
-    iss::status read_dpc_reg(unsigned addr, reg_t& val);
-    iss::status write_dpc_reg(unsigned addr, reg_t val);
+    iss::status write_dcsr(unsigned addr, reg_t val);
+    iss::status read_debug(unsigned addr, reg_t& val);
+    iss::status write_dscratch(unsigned addr, reg_t val);
+    iss::status read_dpc(unsigned addr, reg_t& val);
+    iss::status write_dpc(unsigned addr, reg_t val);
+    iss::status read_fcsr(unsigned addr, reg_t& val);
+    iss::status write_fcsr(unsigned addr, reg_t val);
 
-    virtual iss::status read_custom_csr_reg(unsigned addr, reg_t& val) { return iss::status::Err; };
-    virtual iss::status write_custom_csr_reg(unsigned addr, reg_t val) { return iss::status::Err; };
+    virtual iss::status read_custom_csr(unsigned addr, reg_t& val) { return iss::status::Err; };
+    virtual iss::status write_custom_csr(unsigned addr, reg_t val) { return iss::status::Err; };
 
-    void register_custom_csr_rd(unsigned addr) { csr_rd_cb[addr] = &this_class::read_custom_csr_reg; }
-    void register_custom_csr_wr(unsigned addr) { csr_wr_cb[addr] = &this_class::write_custom_csr_reg; }
+    void register_custom_csr_rd(unsigned addr) { csr_rd_cb[addr] = &this_class::read_custom_csr; }
+    void register_custom_csr_wr(unsigned addr) { csr_wr_cb[addr] = &this_class::write_custom_csr; }
 
     reg_t mhartid_reg{0x0};
 
@@ -445,19 +448,22 @@ riscv_hart_m_p<BASE, FEAT, LOGCAT>::riscv_hart_m_p(feature_config cfg)
     csr[marchid] = traits<BASE>::MARCHID_VAL;
     csr[mimpid] = 1;
 
-    uart_buf.str("");
+    if(traits<BASE>::FLEN > 0) {
+        csr_rd_cb[fcsr] = &this_class::read_fcsr;
+        csr_wr_cb[fcsr] = &this_class::write_fcsr;
+    }
     for(unsigned addr = mhpmcounter3; addr <= mhpmcounter31; ++addr) {
         csr_rd_cb[addr] = &this_class::read_null;
-        csr_wr_cb[addr] = &this_class::write_csr_reg;
+        csr_wr_cb[addr] = &this_class::write_plain;
     }
     if(traits<BASE>::XLEN == 32)
         for(unsigned addr = mhpmcounter3h; addr <= mhpmcounter31h; ++addr) {
             csr_rd_cb[addr] = &this_class::read_null;
-            csr_wr_cb[addr] = &this_class::write_csr_reg;
+            csr_wr_cb[addr] = &this_class::write_plain;
         }
     for(unsigned addr = mhpmevent3; addr <= mhpmevent31; ++addr) {
         csr_rd_cb[addr] = &this_class::read_null;
-        csr_wr_cb[addr] = &this_class::write_csr_reg;
+        csr_wr_cb[addr] = &this_class::write_plain;
     }
     for(unsigned addr = hpmcounter3; addr <= hpmcounter31; ++addr) {
         csr_rd_cb[addr] = &this_class::read_null;
@@ -465,18 +471,17 @@ riscv_hart_m_p<BASE, FEAT, LOGCAT>::riscv_hart_m_p(feature_config cfg)
     if(traits<BASE>::XLEN == 32)
         for(unsigned addr = hpmcounter3h; addr <= hpmcounter31h; ++addr) {
             csr_rd_cb[addr] = &this_class::read_null;
-            // csr_wr_cb[addr] = &this_class::write_csr_reg;
         }
     // common regs
     const std::array<unsigned, 4> roaddrs{{misa, mvendorid, marchid, mimpid}};
     for(auto addr : roaddrs) {
-        csr_rd_cb[addr] = &this_class::read_csr_reg;
+        csr_rd_cb[addr] = &this_class::read_plain;
         csr_wr_cb[addr] = &this_class::write_null;
     }
     const std::array<unsigned, 4> rwaddrs{{mepc, mtvec, mscratch, mtval}};
     for(auto addr : rwaddrs) {
-        csr_rd_cb[addr] = &this_class::read_csr_reg;
-        csr_wr_cb[addr] = &this_class::write_csr_reg;
+        csr_rd_cb[addr] = &this_class::read_plain;
+        csr_wr_cb[addr] = &this_class::write_plain;
     }
     // special handling & overrides
     csr_rd_cb[time] = &this_class::read_time;
@@ -517,7 +522,7 @@ riscv_hart_m_p<BASE, FEAT, LOGCAT>::riscv_hart_m_p(feature_config cfg)
     csr_wr_cb[marchid] = &this_class::write_null;
     csr_wr_cb[mimpid] = &this_class::write_null;
     if(FEAT & FEAT_CLIC) {
-        csr_rd_cb[mtvt] = &this_class::read_csr_reg;
+        csr_rd_cb[mtvt] = &this_class::read_plain;
         csr_wr_cb[mtvt] = &this_class::write_xtvt;
         //        csr_rd_cb[mxnti] = &this_class::read_csr_reg;
         //        csr_wr_cb[mxnti] = &this_class::write_csr_reg;
@@ -527,7 +532,7 @@ riscv_hart_m_p<BASE, FEAT, LOGCAT>::riscv_hart_m_p(feature_config cfg)
         //        csr_wr_cb[mscratchcsw] = &this_class::write_csr_reg;
         //        csr_rd_cb[mscratchcswl] = &this_class::read_csr_reg;
         //        csr_wr_cb[mscratchcswl] = &this_class::write_csr_reg;
-        csr_rd_cb[mintthresh] = &this_class::read_csr_reg;
+        csr_rd_cb[mintthresh] = &this_class::read_plain;
         csr_wr_cb[mintthresh] = &this_class::write_intthresh;
         clic_int_reg.resize(cfg.clic_num_irq, clic_int_reg_t{.raw = 0});
         clic_cfg_reg = 0x20;
@@ -553,14 +558,14 @@ riscv_hart_m_p<BASE, FEAT, LOGCAT>::riscv_hart_m_p(feature_config cfg)
         insert_mem_range(cfg.tcm_base, cfg.tcm_size, read_clic_cb, write_clic_cb);
     }
     if(FEAT & FEAT_DEBUG) {
-        csr_wr_cb[dscratch0] = &this_class::write_dcsr_reg;
-        csr_rd_cb[dscratch0] = &this_class::read_dcsr_reg;
-        csr_wr_cb[dscratch1] = &this_class::write_dcsr_reg;
-        csr_rd_cb[dscratch1] = &this_class::read_dcsr_reg;
-        csr_wr_cb[dpc] = &this_class::write_dpc_reg;
-        csr_rd_cb[dpc] = &this_class::read_dpc_reg;
-        csr_wr_cb[dcsr] = &this_class::write_dcsr_dcsr;
-        csr_rd_cb[dcsr] = &this_class::read_dcsr_reg;
+        csr_wr_cb[dscratch0] = &this_class::write_dscratch;
+        csr_rd_cb[dscratch0] = &this_class::read_debug;
+        csr_wr_cb[dscratch1] = &this_class::write_dscratch;
+        csr_rd_cb[dscratch1] = &this_class::read_debug;
+        csr_wr_cb[dpc] = &this_class::write_dpc;
+        csr_rd_cb[dpc] = &this_class::read_dpc;
+        csr_wr_cb[dcsr] = &this_class::write_dcsr;
+        csr_rd_cb[dcsr] = &this_class::read_debug;
     }
     hart_mem_rd_delegate = [this](phys_addr_t a, unsigned l, uint8_t* const d) -> iss::status { return this->read_mem(a, l, d); };
     hart_mem_wr_delegate = [this](phys_addr_t a, unsigned l, uint8_t const* const d) -> iss::status { return this->write_mem(a, l, d); };
@@ -568,57 +573,14 @@ riscv_hart_m_p<BASE, FEAT, LOGCAT>::riscv_hart_m_p(feature_config cfg)
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
 std::pair<uint64_t, bool> riscv_hart_m_p<BASE, FEAT, LOGCAT>::load_file(std::string name, int type) {
-    get_sym_table(name);
-    try {
-        tohost = symbol_table.at("tohost");
-        fromhost = symbol_table.at("fromhost");
-    } catch(std::out_of_range& e) {
+    if(read_elf_file(name, sizeof(reg_t) == 4 ? ELFIO::ELFCLASS32 : ELFIO::ELFCLASS64,
+                     [this](uint64_t addr, uint64_t size, const uint8_t* const data) -> iss::status {
+                         return this->write(iss::address_type::PHYSICAL, iss::access_type::DEBUG_WRITE, traits<BASE>::MEM, addr, size,
+                                            data);
+                     })) {
+        return std::make_pair(entry_address, true);
     }
-    FILE* fp = fopen(name.c_str(), "r");
-    if(fp) {
-        std::array<char, 5> buf;
-        auto n = fread(buf.data(), 1, 4, fp);
-        fclose(fp);
-        if(n != 4)
-            throw std::runtime_error("input file has insufficient size");
-        buf[4] = 0;
-        if(strcmp(buf.data() + 1, "ELF") == 0) {
-            // Create elfio reader
-            ELFIO::elfio reader;
-            // Load ELF data
-            if(!reader.load(name))
-                throw std::runtime_error("could not process elf file");
-            // check elf properties
-            if(reader.get_class() != ELFCLASS32)
-                if(sizeof(reg_t) == 4)
-                    throw std::runtime_error("wrong elf class in file");
-            if(reader.get_type() != ET_EXEC)
-                throw std::runtime_error("wrong elf type in file");
-            if(reader.get_machine() != EM_RISCV)
-                throw std::runtime_error("wrong elf machine in file");
-            auto entry = reader.get_entry();
-            for(const auto pseg : reader.segments) {
-                const auto fsize = pseg->get_file_size(); // 0x42c/0x0
-                const auto seg_data = pseg->get_data();
-                const auto type = pseg->get_type();
-                if(type == 1 && fsize > 0) {
-                    auto res = this->write(iss::address_type::PHYSICAL, iss::access_type::DEBUG_WRITE, traits<BASE>::MEM,
-                                           pseg->get_physical_address(), fsize, reinterpret_cast<const uint8_t* const>(seg_data));
-                    if(res != iss::Ok)
-                        CPPLOG(ERR) << "problem writing " << fsize << "bytes to 0x" << std::hex << pseg->get_physical_address();
-                }
-            }
-            for(const auto sec : reader.sections) {
-                if(sec->get_name() == ".tohost") {
-                    tohost = sec->get_address();
-                    fromhost = tohost + 0x40;
-                }
-            }
-            return std::make_pair(entry, true);
-        }
-        throw std::runtime_error(fmt::format("memory load file {} is not a valid elf file", name));
-    }
-    throw std::runtime_error(fmt::format("memory load file not found, check if {} is a valid file", name));
+    return std::make_pair(entry_address, false);
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
@@ -649,7 +611,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read(const address_type type, co
     try {
         switch(space) {
         case traits<BASE>::MEM: {
-            auto alignment = is_fetch(access) ? (has_compressed() ? 2 : 4) : length;
+            auto alignment = is_fetch(access) ? (has_compressed() ? 2 : 4) : std::min<unsigned>(length, sizeof(reg_t));
             if(unlikely(is_fetch(access) && (addr & (alignment - 1)))) {
                 fault_data = addr;
                 if(is_debug(access))
@@ -665,7 +627,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read(const address_type type, co
                 }
                 phys_addr_t phys_addr{access, space, addr};
                 auto res = iss::Err;
-                if(access != access_type::FETCH && memfn_range.size()) {
+                if(!is_fetch(access) && memfn_range.size()) {
                     auto it =
                         std::find_if(std::begin(memfn_range), std::end(memfn_range), [phys_addr](std::tuple<uint64_t, uint64_t> const& a) {
                             return std::get<0>(a) <= phys_addr.val && (std::get<0>(a) + std::get<1>(a)) > phys_addr.val;
@@ -684,19 +646,16 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read(const address_type type, co
                 }
                 return res;
             } catch(trap_access& ta) {
-                this->reg.trap_state = (1UL << 31) | ta.id;
-                fault_data = ta.addr;
+                if((access & access_type::DEBUG) == 0) {
+                    this->reg.trap_state = (1UL << 31) | ta.id;
+                    fault_data = ta.addr;
+                }
                 return iss::Err;
             }
         } break;
         case traits<BASE>::CSR: {
             if(length != sizeof(reg_t))
                 return iss::Err;
-            // We emulate the FCSR in the architectural state
-            if(addr == 3) {
-                *data = this->get_fcsr();
-                return iss::Ok;
-            }
             return read_csr(addr, *reinterpret_cast<reg_t* const>(data));
         } break;
         case traits<BASE>::FENCE: {
@@ -717,8 +676,10 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read(const address_type type, co
         }
         return iss::Ok;
     } catch(trap_access& ta) {
-        this->reg.trap_state = (1UL << 31) | ta.id;
-        fault_data = ta.addr;
+        if((access & access_type::DEBUG) == 0) {
+            this->reg.trap_state = (1UL << 31) | ta.id;
+            fault_data = ta.addr;
+        }
         return iss::Err;
     }
 }
@@ -746,7 +707,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write(const address_type type, c
                       << std::hex << addr;
         break;
     default:
-        CPPLOG(TRACE) << prefix << "write of " << length << " bytes @addr " << addr;
+        CPPLOG(TRACE) << prefix << "write of " << length << " bytes @addr 0x" << std::hex << addr;
     }
 #endif
     try {
@@ -760,14 +721,15 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write(const address_type type, c
                 return iss::Err;
             }
             try {
-                if(length > 1 && (addr & (length - 1)) && (access & access_type::DEBUG) != access_type::DEBUG) {
+                auto alignment = std::min<unsigned>(length, sizeof(reg_t));
+                if(length > 1 && (addr & (alignment - 1)) && !is_debug(access)) {
                     this->reg.trap_state = (1UL << 31) | 6 << 16;
                     fault_data = addr;
                     return iss::Err;
                 }
                 phys_addr_t phys_addr{access, space, addr};
                 auto res = iss::Err;
-                if(access != access_type::FETCH && memfn_range.size()) {
+                if(!is_fetch(access) && memfn_range.size()) {
                     auto it =
                         std::find_if(std::begin(memfn_range), std::end(memfn_range), [phys_addr](std::tuple<uint64_t, uint64_t> const& a) {
                             return std::get<0>(a) <= phys_addr.val && (std::get<0>(a) + std::get<1>(a)) > phys_addr.val;
@@ -780,7 +742,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write(const address_type type, c
                 } else {
                     res = write_mem(phys_addr, length, data);
                 }
-                if(unlikely(res != iss::Ok && (access & access_type::DEBUG) == 0)) {
+                if(unlikely(res != iss::Ok && !is_debug(access))) {
                     this->reg.trap_state = (1UL << 31) | (7UL << 16); // issue trap 7 (Store/AMO access fault)
                     fault_data = addr;
                 }
@@ -790,49 +752,10 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write(const address_type type, c
                 fault_data = ta.addr;
                 return iss::Err;
             }
-
-            if((addr + length) > mem.size())
-                return iss::Err;
-            switch(addr) {
-            case 0x10013000: // UART0 base, TXFIFO reg
-            case 0x10023000: // UART1 base, TXFIFO reg
-                uart_buf << (char)data[0];
-                if(((char)data[0]) == '\n' || data[0] == 0) {
-                    // CPPLOG(INFO)<<"UART"<<((paddr.val>>16)&0x3)<<" send
-                    // '"<<uart_buf.str()<<"'";
-                    std::cout << uart_buf.str();
-                    uart_buf.str("");
-                }
-                return iss::Ok;
-            case 0x10008000: { // HFROSC base, hfrosccfg reg
-                auto& p = mem(addr / mem.page_size);
-                auto offs = addr & mem.page_addr_mask;
-                std::copy(data, data + length, p.data() + offs);
-                auto& x = *(p.data() + offs + 3);
-                if(x & 0x40)
-                    x |= 0x80; // hfroscrdy = 1 if hfroscen==1
-                return iss::Ok;
-            }
-            case 0x10008008: { // HFROSC base, pllcfg reg
-                auto& p = mem(addr / mem.page_size);
-                auto offs = addr & mem.page_addr_mask;
-                std::copy(data, data + length, p.data() + offs);
-                auto& x = *(p.data() + offs + 3);
-                x |= 0x80; // set pll lock upon writing
-                return iss::Ok;
-            } break;
-            default: {
-            }
-            }
         } break;
         case traits<BASE>::CSR: {
             if(length != sizeof(reg_t))
                 return iss::Err;
-            // We emulate the FCSR in the architectural state
-            if(addr == 3) {
-                this->set_fcsr(*data);
-                return iss::Ok;
-            }
             return write_csr(addr, *reinterpret_cast<const reg_t*>(data));
         } break;
         case traits<BASE>::FENCE: {
@@ -855,8 +778,10 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write(const address_type type, c
         }
         return iss::Ok;
     } catch(trap_access& ta) {
-        this->reg.trap_state = (1UL << 31) | ta.id;
-        fault_data = ta.addr;
+        if((access & access_type::DEBUG) == 0) {
+            this->reg.trap_state = (1UL << 31) | ta.id;
+            fault_data = ta.addr;
+        }
         return iss::Err;
     }
 }
@@ -889,12 +814,6 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_csr(unsigned addr, reg_t v
     return (this->*(it->second))(addr, val);
 }
 
-template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_csr_reg(unsigned addr, reg_t& val) {
-    val = csr[addr];
-    return iss::Ok;
-}
-
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_null(unsigned addr, reg_t& val) {
     val = 0;
@@ -902,14 +821,20 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_null(unsigned addr, reg_t&
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_csr_reg(unsigned addr, reg_t val) {
+iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_plain(unsigned addr, reg_t& val) {
+    val = csr[addr];
+    return iss::Ok;
+}
+
+template <typename BASE, features_e FEAT, typename LOGCAT>
+iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_plain(unsigned addr, reg_t val) {
     csr[addr] = val;
     return iss::Ok;
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_cycle(unsigned addr, reg_t& val) {
-    auto cycle_val = this->reg.icount + cycle_offset;
+    auto cycle_val = this->reg.cycle + cycle_offset;
     if(addr == mcycle) {
         val = static_cast<reg_t>(cycle_val);
     } else if(addr == mcycleh) {
@@ -929,7 +854,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_cycle(unsigned addr, reg_t
             mcycle_csr = (static_cast<uint64_t>(val) << 32) + (mcycle_csr & 0xffffffff);
         }
     }
-    cycle_offset = mcycle_csr - this->reg.icount; // TODO: relying on wrap-around
+    cycle_offset = mcycle_csr - this->reg.cycle; // TODO: relying on wrap-around
     return iss::Ok;
 }
 
@@ -960,7 +885,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_instret(unsigned addr, reg
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_time(unsigned addr, reg_t& val) {
-    uint64_t time_val = this->reg.icount / (100000000 / 32768 - 1); //-> ~3052;
+    uint64_t time_val = this->reg.cycle / (100000000 / 32768 - 1); //-> ~3052;
     if(addr == time) {
         val = static_cast<reg_t>(time_val);
     } else if(addr == timeh) {
@@ -1052,7 +977,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_epc(unsigned addr, reg_t v
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_dcsr_dcsr(unsigned addr, reg_t val) {
+iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_dcsr(unsigned addr, reg_t val) {
     if(!debug_mode_active())
         throw illegal_instruction_fault(this->fault_data);
     //                  +-------------- ebreakm
@@ -1064,7 +989,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_dcsr_dcsr(unsigned addr, r
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_dcsr_reg(unsigned addr, reg_t& val) {
+iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_debug(unsigned addr, reg_t& val) {
     if(!debug_mode_active())
         throw illegal_instruction_fault(this->fault_data);
     val = csr[addr];
@@ -1072,7 +997,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_dcsr_reg(unsigned addr, reg
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_dcsr_reg(unsigned addr, reg_t val) {
+iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_dscratch(unsigned addr, reg_t val) {
     if(!debug_mode_active())
         throw illegal_instruction_fault(this->fault_data);
     csr[addr] = val;
@@ -1080,7 +1005,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_dcsr_reg(unsigned addr, re
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_dpc_reg(unsigned addr, reg_t& val) {
+iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_dpc(unsigned addr, reg_t& val) {
     if(!debug_mode_active())
         throw illegal_instruction_fault(this->fault_data);
     val = this->reg.DPC;
@@ -1088,7 +1013,7 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_dpc_reg(unsigned addr, reg_
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_dpc_reg(unsigned addr, reg_t val) {
+iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_dpc(unsigned addr, reg_t val) {
     if(!debug_mode_active())
         throw illegal_instruction_fault(this->fault_data);
     this->reg.DPC = val;
@@ -1101,6 +1026,18 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_intstatus(unsigned addr, re
     return iss::Ok;
 }
 
+template <typename BASE, features_e FEAT, typename LOGCAT>
+iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_fcsr(unsigned addr, reg_t& val) {
+    val = this->get_fcsr();
+    return iss::Ok;
+}
+
+template <typename BASE, features_e FEAT, typename LOGCAT>
+iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_fcsr(unsigned addr, reg_t val) {
+    this->set_fcsr(val);
+    return iss::Ok;
+}
+
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_intthresh(unsigned addr, reg_t val) {
     csr[addr] = (val & 0xff) | (1 << (cfg.clic_int_ctl_bits)) - 1;
@@ -1127,59 +1064,51 @@ iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::read_mem(phys_addr_t paddr, unsi
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_m_p<BASE, FEAT, LOGCAT>::write_mem(phys_addr_t paddr, unsigned length, const uint8_t* const data) {
-    switch(paddr.val) {
-    // TODO remove UART, Peripherals should not be part of the ISS
-    case 0xFFFF0000: // UART0 base, TXFIFO reg
-        if(((char)data[0]) == '\n' || data[0] == 0) {
-            CPPLOG(INFO) << "UART" << ((paddr.val >> 12) & 0x3) << " send '" << uart_buf.str() << "'";
-            uart_buf.str("");
-        } else if(((char)data[0]) != '\r')
-            uart_buf << (char)data[0];
-        break;
-    default: {
-        mem_type::page_type& p = mem(paddr.val / mem.page_size);
-        std::copy(data, data + length, p.data() + (paddr.val & mem.page_addr_mask));
-        // tohost handling in case of riscv-test
-        if(paddr.access && iss::access_type::FUNC) {
-            auto tohost_upper =
-                (traits<BASE>::XLEN == 32 && paddr.val == (tohost + 4)) || (traits<BASE>::XLEN == 64 && paddr.val == tohost);
-            auto tohost_lower = (traits<BASE>::XLEN == 32 && paddr.val == tohost) || (traits<BASE>::XLEN == 64 && paddr.val == tohost);
-            if(tohost_lower || tohost_upper) {
-                uint64_t hostvar = *reinterpret_cast<uint64_t*>(p.data() + (tohost & mem.page_addr_mask));
-                // in case of 32 bit system, two writes to tohost are needed, only evaluate on the second (high) write
-                if(tohost_upper && (tohost_lower || tohost_lower_written)) {
-                    switch(hostvar >> 48) {
-                    case 0:
-                        if(hostvar != 0x1) {
-                            CPPLOG(FATAL) << "tohost value is 0x" << std::hex << hostvar << std::dec << " (" << hostvar
-                                          << "), stopping simulation";
-                        } else {
-                            CPPLOG(INFO) << "tohost value is 0x" << std::hex << hostvar << std::dec << " (" << hostvar
-                                         << "), stopping simulation";
-                        }
-                        this->reg.trap_state = std::numeric_limits<uint32_t>::max();
-                        this->interrupt_sim = hostvar;
-                        break;
-                    case 0x0101: {
-                        char c = static_cast<char>(hostvar & 0xff);
-                        if(c == '\n' || c == 0) {
-                            CPPLOG(INFO) << "tohost send '" << uart_buf.str() << "'";
-                            uart_buf.str("");
-                        } else
-                            uart_buf << c;
-                    } break;
-                    default:
-                        break;
-                    }
-                    tohost_lower_written = false;
-                } else if(tohost_lower)
-                    tohost_lower_written = true;
-            } else if((traits<BASE>::XLEN == 32 && paddr.val == fromhost + 4) || (traits<BASE>::XLEN == 64 && paddr.val == fromhost)) {
-                uint64_t fhostvar = *reinterpret_cast<uint64_t*>(p.data() + (fromhost & mem.page_addr_mask));
-                *reinterpret_cast<uint64_t*>(p.data() + (tohost & mem.page_addr_mask)) = fhostvar;
+    mem_type::page_type& p = mem(paddr.val / mem.page_size);
+    std::copy(data, data + length, p.data() + (paddr.val & mem.page_addr_mask));
+    // tohost handling in case of riscv-test
+    // according to https://github.com/riscv-software-src/riscv-isa-sim/issues/364#issuecomment-607657754:
+    if(paddr.access && iss::access_type::FUNC) {
+        if(paddr.val == tohost) {
+            reg_t cur_data = *reinterpret_cast<const reg_t*>(data);
+            // Extract Device (bits 63:56)
+            uint8_t device = traits<BASE>::XLEN == 32 ? 0 : (cur_data >> 56) & 0xFF;
+            // Extract Command (bits 55:48)
+            uint8_t command = traits<BASE>::XLEN == 32 ? 0 : (cur_data >> 48) & 0xFF;
+            // Extract payload (bits 47:0)
+            uint64_t payload_addr = cur_data & 0xFFFFFFFFFFFFULL;
+            if(payload_addr & 1) {
+                CPPLOG(FATAL) << "tohost value is 0x" << std::hex << payload_addr << std::dec << " (" << payload_addr
+                              << "), stopping simulation";
+                this->reg.trap_state = std::numeric_limits<uint32_t>::max();
+                this->interrupt_sim = payload_addr;
+                return iss::Ok;
+            } else if(device == 0 && command == 0) {
+                std::array<uint64_t, 8> loaded_payload;
+                if(read(address_type::PHYSICAL, access_type::DEBUG_READ, traits<BASE>::MEM, payload_addr, 8 * sizeof(uint64_t),
+                        reinterpret_cast<uint8_t*>(loaded_payload.data())) == iss::Err)
+                    CPPLOG(ERR) << "Syscall read went wrong";
+                uint64_t syscall_num = loaded_payload.at(0);
+                if(syscall_num == 64) { // SYS_WRITE
+                    return execute_sys_write(this, loaded_payload, traits<BASE>::MEM);
+                } else {
+                    CPPLOG(ERR) << "tohost syscall with number 0x" << std::hex << syscall_num << std::dec << " (" << syscall_num
+                                << ") not implemented";
+                    this->reg.trap_state = std::numeric_limits<uint32_t>::max();
+                    this->interrupt_sim = payload_addr;
+                    return iss::Ok;
+                }
+            } else {
+                CPPLOG(ERR) << "tohost functionality not implemented for device " << device << " and command " << command;
+                this->reg.trap_state = std::numeric_limits<uint32_t>::max();
+                this->interrupt_sim = payload_addr;
+                return iss::Ok;
             }
         }
-    }
+        if((traits<BASE>::XLEN == 32 && paddr.val == fromhost + 4) || (traits<BASE>::XLEN == 64 && paddr.val == fromhost)) {
+            uint64_t fhostvar = *reinterpret_cast<uint64_t*>(p.data() + (fromhost & mem.page_addr_mask));
+            *reinterpret_cast<uint64_t*>(p.data() + (tohost & mem.page_addr_mask)) = fhostvar;
+        }
     }
     return iss::Ok;
 }

src/iss/arch/riscv_hart_msu_vp.h

@@ -39,7 +39,14 @@
 #include "iss/instrumentation_if.h"
 #include "iss/log_categories.h"
 #include "iss/vm_if.h"
+#include "iss/vm_types.h"
 #include "riscv_hart_common.h"
+#include "util/logging.h"
+#include <algorithm>
+#include <cstdint>
+#include <elfio/elf_types.hpp>
+#include <limits>
+#include <stdexcept>
 #ifndef FMT_HEADER_ONLY
 #define FMT_HEADER_ONLY
 #endif
@@ -326,7 +333,7 @@ public:
 
     void disass_output(uint64_t pc, const std::string instr) override {
         CLOG(INFO, disass) << fmt::format("0x{:016x}    {:40} [p:{};s:0x{:x};c:{}]", pc, instr, lvl[this->reg.PRIV], (reg_t)state.mstatus,
-                                          this->reg.icount + cycle_offset);
+                                          this->reg.cycle + cycle_offset);
     };
 
     iss::instrumentation_if* get_instrumentation_if() override { return &instr_if; }
@@ -359,7 +366,7 @@ protected:
 
         uint64_t get_pendig_traps() override { return arch.reg.trap_state; }
 
-        uint64_t get_total_cycles() override { return arch.reg.icount + arch.cycle_offset; }
+        uint64_t get_total_cycles() override { return arch.reg.cycle + arch.cycle_offset; }
 
         void update_last_instr_cycles(unsigned cycles) override { arch.cycle_offset += cycles - 1; }
 
@@ -369,7 +376,7 @@ protected:
 
         unsigned get_reg_size(unsigned num) override { return traits<BASE>::reg_bit_widths[num]; }
 
-        std::unordered_map<std::string, uint64_t> get_symbol_table(std::string name) override { return arch.get_sym_table(name); }
+        std::unordered_map<std::string, uint64_t> const& get_symbol_table(std::string name) override { return arch.symbol_table; }
 
         riscv_hart_msu_vp<BASE>& arch;
     };
@@ -391,9 +398,6 @@ protected:
     uint64_t minstret_csr{0};
     reg_t fault_data;
     std::array<vm_info, 2> vm;
-    uint64_t tohost = tohost_dflt;
-    uint64_t fromhost = fromhost_dflt;
-    bool tohost_lower_written = false;
     riscv_instrumentation_if instr_if;
 
     std::function<void(arch_if*, reg_t, reg_t)> semihosting_cb;
@@ -404,7 +408,6 @@ protected:
     mem_type mem;
     csr_type csr;
     void update_vm_info();
-    std::stringstream uart_buf;
     std::unordered_map<reg_t, uint64_t> ptw;
     std::unordered_map<uint64_t, uint8_t> atomic_reservation;
     std::unordered_map<unsigned, rd_csr_f> csr_rd_cb;
@@ -459,7 +462,6 @@ riscv_hart_msu_vp<BASE>::riscv_hart_msu_vp()
     csr[marchid] = traits<BASE>::MARCHID_VAL;
     csr[mimpid] = 1;
 
-    uart_buf.str("");
     for(unsigned addr = mhpmcounter3; addr <= mhpmcounter31; ++addr) {
         csr_rd_cb[addr] = &this_class::read_null;
         csr_wr_cb[addr] = &this_class::write_csr_reg;
@@ -555,71 +557,14 @@ riscv_hart_msu_vp<BASE>::riscv_hart_msu_vp()
 }
 
 template <typename BASE> std::pair<uint64_t, bool> riscv_hart_msu_vp<BASE>::load_file(std::string name, int type) {
-    FILE* fp = fopen(name.c_str(), "r");
-    if(fp) {
-        std::array<char, 5> buf;
-        auto n = fread(buf.data(), 1, 4, fp);
-        fclose(fp);
-        if(n != 4)
-            throw std::runtime_error("input file has insufficient size");
-        buf[4] = 0;
-        if(strcmp(buf.data() + 1, "ELF") == 0) {
-            // Create elfio reader
-            ELFIO::elfio reader;
-            // Load ELF data
-            if(!reader.load(name))
-                throw std::runtime_error("could not process elf file");
-            // check elf properties
-            if(reader.get_class() != ELFCLASS32)
-                if(sizeof(reg_t) == 4)
-                    throw std::runtime_error("wrong elf class in file");
-            if(reader.get_type() != ET_EXEC)
-                throw std::runtime_error("wrong elf type in file");
-            if(reader.get_machine() != EM_RISCV)
-                throw std::runtime_error("wrong elf machine in file");
-            auto entry = reader.get_entry();
-            for(const auto pseg : reader.segments) {
-                const auto fsize = pseg->get_file_size(); // 0x42c/0x0
-                const auto seg_data = pseg->get_data();
-                const auto type = pseg->get_type();
-                if(type == 1 && fsize > 0) {
-                    auto res = this->write(iss::address_type::PHYSICAL, iss::access_type::DEBUG_WRITE, traits<BASE>::MEM,
-                                           pseg->get_physical_address(), fsize, reinterpret_cast<const uint8_t* const>(seg_data));
-                    if(res != iss::Ok)
-                        CPPLOG(ERR) << "problem writing " << fsize << "bytes to 0x" << std::hex << pseg->get_physical_address();
-                }
-            }
-            for(const auto sec : reader.sections) {
-                if(sec->get_name() == ".symtab") {
-                    if(SHT_SYMTAB == sec->get_type() || SHT_DYNSYM == sec->get_type()) {
-                        ELFIO::symbol_section_accessor symbols(reader, sec);
-                        auto sym_no = symbols.get_symbols_num();
-                        std::string name;
-                        ELFIO::Elf64_Addr value = 0;
-                        ELFIO::Elf_Xword size = 0;
-                        unsigned char bind = 0;
-                        unsigned char type = 0;
-                        ELFIO::Elf_Half section = 0;
-                        unsigned char other = 0;
-                        for(auto i = 0U; i < sym_no; ++i) {
-                            symbols.get_symbol(i, name, value, size, bind, type, section, other);
-                            if(name == "tohost") {
-                                tohost = value;
-                            } else if(name == "fromhost") {
-                                fromhost = value;
-                            }
-                        }
-                    }
-                } else if(sec->get_name() == ".tohost") {
-                    tohost = sec->get_address();
-                    fromhost = tohost + 0x40;
-                }
-            }
-            return std::make_pair(entry, true);
-        }
-        throw std::runtime_error(fmt::format("memory load file {} is not a valid elf file", name));
+    if(read_elf_file(name, sizeof(reg_t) == 4 ? ELFIO::ELFCLASS32 : ELFIO::ELFCLASS64,
+                     [this](uint64_t addr, uint64_t size, const uint8_t* const data) -> iss::status {
+                         return this->write(iss::address_type::PHYSICAL, iss::access_type::DEBUG_WRITE, traits<BASE>::MEM, addr, size,
+                                            data);
+                     })) {
+        return std::make_pair(entry_address, true);
     }
-    throw std::runtime_error(fmt::format("memory load file not found, check if {} is a valid file", name));
+    return std::make_pair(entry_address, false);
 }
 
 template <typename BASE>
@@ -637,7 +582,7 @@ iss::status riscv_hart_msu_vp<BASE>::read(const address_type type, const access_
     try {
         switch(space) {
         case traits<BASE>::MEM: {
-            auto alignment = is_fetch(access) ? (traits<BASE>::MISA_VAL & 0x100 ? 2 : 4) : length;
+            auto alignment = is_fetch(access) ? (has_compressed() ? 2 : 4) : std::min<unsigned>(length, sizeof(reg_t));
             if(unlikely(is_fetch(access) && (addr & (alignment - 1)))) {
                 fault_data = addr;
                 if(access && iss::access_type::DEBUG)
@@ -669,8 +614,10 @@ iss::status riscv_hart_msu_vp<BASE>::read(const address_type type, const access_
                 }
                 return res;
             } catch(trap_access& ta) {
-                this->reg.trap_state = (1 << 31) | ta.id;
-                fault_data = ta.addr;
+                if((access & access_type::DEBUG) == 0) {
+                    this->reg.trap_state = (1UL << 31) | ta.id;
+                    fault_data = ta.addr;
+                }
                 return iss::Err;
             }
         } break;
@@ -708,8 +655,10 @@ iss::status riscv_hart_msu_vp<BASE>::read(const address_type type, const access_
         }
         return iss::Ok;
     } catch(trap_access& ta) {
-        this->reg.trap_state = (1UL << 31) | ta.id;
-        fault_data = ta.addr;
+        if((access & access_type::DEBUG) == 0) {
+            this->reg.trap_state = (1UL << 31) | ta.id;
+            fault_data = ta.addr;
+        }
         return iss::Err;
     }
 }
@@ -752,6 +701,7 @@ iss::status riscv_hart_msu_vp<BASE>::write(const address_type type, const access
             }
             phys_addr_t paddr = BASE::v2p(iss::addr_t{access, type, space, addr});
             try {
+                // TODO: There is no check for alignment
                 if(unlikely((addr & ~PGMASK) != ((addr + length - 1) & ~PGMASK))) { // we may cross a page boundary
                     vm_info vm = hart_state_type::decode_vm_info(this->reg.PRIV, state.satp);
                     if(vm.levels != 0) { // VM is active
@@ -774,40 +724,6 @@ iss::status riscv_hart_msu_vp<BASE>::write(const address_type type, const access
                 fault_data = ta.addr;
                 return iss::Err;
             }
-
-            if((paddr.val + length) > mem.size())
-                return iss::Err;
-            switch(paddr.val) {
-            case 0x10013000: // UART0 base, TXFIFO reg
-            case 0x10023000: // UART1 base, TXFIFO reg
-                uart_buf << (char)data[0];
-                if(((char)data[0]) == '\n' || data[0] == 0) {
-                    // CPPLOG(INFO)<<"UART"<<((paddr.val>>16)&0x3)<<" send
-                    // '"<<uart_buf.str()<<"'";
-                    std::cout << uart_buf.str();
-                    uart_buf.str("");
-                }
-                return iss::Ok;
-            case 0x10008000: { // HFROSC base, hfrosccfg reg
-                auto& p = mem(paddr.val / mem.page_size);
-                auto offs = paddr.val & mem.page_addr_mask;
-                std::copy(data, data + length, p.data() + offs);
-                auto& x = *(p.data() + offs + 3);
-                if(x & 0x40)
-                    x |= 0x80; // hfroscrdy = 1 if hfroscen==1
-                return iss::Ok;
-            }
-            case 0x10008008: { // HFROSC base, pllcfg reg
-                auto& p = mem(paddr.val / mem.page_size);
-                auto offs = paddr.val & mem.page_addr_mask;
-                std::copy(data, data + length, p.data() + offs);
-                auto& x = *(p.data() + offs + 3);
-                x |= 0x80; // set pll lock upon writing
-                return iss::Ok;
-            } break;
-            default: {
-            }
-            }
         } break;
         case traits<BASE>::CSR: {
             if(length != sizeof(reg_t))
@@ -839,8 +755,10 @@ iss::status riscv_hart_msu_vp<BASE>::write(const address_type type, const access
         }
         return iss::Ok;
     } catch(trap_access& ta) {
-        this->reg.trap_state = (1UL << 31) | ta.id;
-        fault_data = ta.addr;
+        if((access & access_type::DEBUG) == 0) {
+            this->reg.trap_state = (1UL << 31) | ta.id;
+            fault_data = ta.addr;
+        }
         return iss::Err;
     }
 }
@@ -887,7 +805,7 @@ template <typename BASE> iss::status riscv_hart_msu_vp<BASE>::write_reg(unsigned
 }
 
 template <typename BASE> iss::status riscv_hart_msu_vp<BASE>::read_cycle(unsigned addr, reg_t& val) {
-    auto cycle_val = this->reg.icount + cycle_offset;
+    auto cycle_val = this->reg.cycle + cycle_offset;
     if(addr == mcycle) {
         val = static_cast<reg_t>(cycle_val);
     } else if(addr == mcycleh) {
@@ -908,7 +826,7 @@ template <typename BASE> iss::status riscv_hart_msu_vp<BASE>::write_cycle(unsign
             mcycle_csr = (static_cast<uint64_t>(val) << 32) + (mcycle_csr & 0xffffffff);
         }
     }
-    cycle_offset = mcycle_csr - this->reg.icount; // TODO: relying on wrap-around
+    cycle_offset = mcycle_csr - this->reg.cycle; // TODO: relying on wrap-around
     return iss::Ok;
 }
 
@@ -936,7 +854,7 @@ template <typename BASE> iss::status riscv_hart_msu_vp<BASE>::write_instret(unsi
 }
 
 template <typename BASE> iss::status riscv_hart_msu_vp<BASE>::read_time(unsigned addr, reg_t& val) {
-    uint64_t time_val = this->reg.icount / (100000000 / 32768 - 1); //-> ~3052;
+    uint64_t time_val = this->reg.cycle / (100000000 / 32768 - 1); //-> ~3052;
     if(addr == time) {
         val = static_cast<reg_t>(time_val);
     } else if(addr == timeh) {
@@ -1075,61 +993,51 @@ template <typename BASE> iss::status riscv_hart_msu_vp<BASE>::read_mem(phys_addr
 }
 
 template <typename BASE> iss::status riscv_hart_msu_vp<BASE>::write_mem(phys_addr_t paddr, unsigned length, const uint8_t* const data) {
-    switch(paddr.val) {
-    case 0xFFFF0000: // UART0 base, TXFIFO reg
-        if(((char)data[0]) == '\n' || data[0] == 0) {
-            CPPLOG(INFO) << "UART" << ((paddr.val >> 12) & 0x3) << " send '" << uart_buf.str() << "'";
-            uart_buf.str("");
-        } else if(((char)data[0]) != '\r')
-            uart_buf << (char)data[0];
-        break;
-    default: {
-        mem_type::page_type& p = mem(paddr.val / mem.page_size);
-        std::copy(data, data + length, p.data() + (paddr.val & mem.page_addr_mask));
-        // tohost handling in case of riscv-test
-        if(paddr.access && iss::access_type::FUNC) {
-            auto tohost_upper =
-                (traits<BASE>::XLEN == 32 && paddr.val == (tohost + 4)) || (traits<BASE>::XLEN == 64 && paddr.val == tohost);
-            auto tohost_lower = (traits<BASE>::XLEN == 32 && paddr.val == tohost) || (traits<BASE>::XLEN == 64 && paddr.val == tohost);
-            if(tohost_lower || tohost_upper) {
-                uint64_t hostvar = *reinterpret_cast<uint64_t*>(p.data() + (tohost & mem.page_addr_mask));
-                // in case of 32 bit system, two writes to tohost are needed, only evaluate on the second (high) write
-                if(tohost_upper && (tohost_lower || tohost_lower_written)) {
-                    switch(hostvar >> 48) {
-                    case 0:
-                        if(hostvar != 0x1) {
-                            CPPLOG(FATAL) << "tohost value is 0x" << std::hex << hostvar << std::dec << " (" << hostvar
-                                          << "), stopping simulation";
-                        } else {
-                            CPPLOG(INFO) << "tohost value is 0x" << std::hex << hostvar << std::dec << " (" << hostvar
-                                         << "), stopping simulation";
-                        }
-                        this->reg.trap_state = std::numeric_limits<uint32_t>::max();
-                        this->interrupt_sim = hostvar;
-#ifndef WITH_TCC
-                        throw(iss::simulation_stopped(hostvar));
-#endif
-                        break;
-                    case 0x0101: {
-                        char c = static_cast<char>(hostvar & 0xff);
-                        if(c == '\n' || c == 0) {
-                            CPPLOG(INFO) << "tohost send '" << uart_buf.str() << "'";
-                            uart_buf.str("");
-                        } else
-                            uart_buf << c;
-                    } break;
-                    default:
-                        break;
-                    }
-                    tohost_lower_written = false;
-                } else if(tohost_lower)
-                    tohost_lower_written = true;
-            } else if((traits<BASE>::XLEN == 32 && paddr.val == fromhost + 4) || (traits<BASE>::XLEN == 64 && paddr.val == fromhost)) {
-                uint64_t fhostvar = *reinterpret_cast<uint64_t*>(p.data() + (fromhost & mem.page_addr_mask));
-                *reinterpret_cast<uint64_t*>(p.data() + (tohost & mem.page_addr_mask)) = fhostvar;
+    mem_type::page_type& p = mem(paddr.val / mem.page_size);
+    std::copy(data, data + length, p.data() + (paddr.val & mem.page_addr_mask));
+    // tohost handling in case of riscv-test
+    // according to https://github.com/riscv-software-src/riscv-isa-sim/issues/364#issuecomment-607657754:
+    if(paddr.access && iss::access_type::FUNC) {
+        if(paddr.val == tohost) {
+            reg_t cur_data = *reinterpret_cast<const reg_t*>(data);
+            // Extract Device (bits 63:56)
+            uint8_t device = traits<BASE>::XLEN == 32 ? 0 : (cur_data >> 56) & 0xFF;
+            // Extract Command (bits 55:48)
+            uint8_t command = traits<BASE>::XLEN == 32 ? 0 : (cur_data >> 48) & 0xFF;
+            // Extract payload (bits 47:0)
+            uint64_t payload_addr = cur_data & 0xFFFFFFFFFFFFULL;
+            if(payload_addr & 1) {
+                CPPLOG(FATAL) << "tohost value is 0x" << std::hex << payload_addr << std::dec << " (" << payload_addr
+                              << "), stopping simulation";
+                this->reg.trap_state = std::numeric_limits<uint32_t>::max();
+                this->interrupt_sim = payload_addr;
+                return iss::Ok;
+            } else if(device == 0 && command == 0) {
+                std::array<uint64_t, 8> loaded_payload;
+                if(read(address_type::PHYSICAL, access_type::DEBUG_READ, traits<BASE>::MEM, payload_addr, 8 * sizeof(uint64_t),
+                        reinterpret_cast<uint8_t*>(loaded_payload.data())) == iss::Err)
+                    CPPLOG(ERR) << "Syscall read went wrong";
+                uint64_t syscall_num = loaded_payload.at(0);
+                if(syscall_num == 64) { // SYS_WRITE
+                    return execute_sys_write(this, loaded_payload, traits<BASE>::MEM);
+                } else {
+                    CPPLOG(ERR) << "tohost syscall with number 0x" << std::hex << syscall_num << std::dec << " (" << syscall_num
+                                << ") not implemented";
+                    this->reg.trap_state = std::numeric_limits<uint32_t>::max();
+                    this->interrupt_sim = payload_addr;
+                    return iss::Ok;
+                }
+            } else {
+                CPPLOG(ERR) << "tohost functionality not implemented for device " << device << " and command " << command;
+                this->reg.trap_state = std::numeric_limits<uint32_t>::max();
+                this->interrupt_sim = payload_addr;
+                return iss::Ok;
             }
         }
-    }
+        if((traits<BASE>::XLEN == 32 && paddr.val == fromhost + 4) || (traits<BASE>::XLEN == 64 && paddr.val == fromhost)) {
+            uint64_t fhostvar = *reinterpret_cast<uint64_t*>(p.data() + (fromhost & mem.page_addr_mask));
+            *reinterpret_cast<uint64_t*>(p.data() + (tohost & mem.page_addr_mask)) = fhostvar;
+        }
     }
     return iss::Ok;
 }

src/iss/arch/riscv_hart_mu_p.h

@@ -39,7 +39,14 @@
 #include "iss/instrumentation_if.h"
 #include "iss/log_categories.h"
 #include "iss/vm_if.h"
+#include "iss/vm_types.h"
 #include "riscv_hart_common.h"
+#include "util/logging.h"
+#include <algorithm>
+#include <cstdint>
+#include <elfio/elf_types.hpp>
+#include <limits>
+#include <stdexcept>
 #ifndef FMT_HEADER_ONLY
 #define FMT_HEADER_ONLY
 #endif
@@ -302,8 +309,8 @@ public:
     void set_mhartid(reg_t mhartid) { mhartid_reg = mhartid; };
 
     void disass_output(uint64_t pc, const std::string instr) override {
-        CLOG(INFO, disass) << fmt::format("0x{:016x}    {:40} [p:{};s:0x{:x};c:{}]", pc, instr, lvl[this->reg.PRIV], (reg_t)state.mstatus,
-                                          this->reg.icount + cycle_offset);
+        NSCLOG(INFO, LOGCAT) << fmt::format("0x{:016x}    {:40} [p:{};s:0x{:x};c:{}]", pc, instr, lvl[this->reg.PRIV], (reg_t)state.mstatus,
+                                            this->reg.cycle + cycle_offset);
     };
 
     iss::instrumentation_if* get_instrumentation_if() override { return &instr_if; }
@@ -336,7 +343,7 @@ protected:
 
         uint64_t get_pendig_traps() override { return arch.reg.trap_state; }
 
-        uint64_t get_total_cycles() override { return arch.reg.icount + arch.cycle_offset; }
+        uint64_t get_total_cycles() override { return arch.reg.cycle + arch.cycle_offset; }
 
         void update_last_instr_cycles(unsigned cycles) override { arch.cycle_offset += cycles - 1; }
 
@@ -346,7 +353,7 @@ protected:
 
         unsigned get_reg_size(unsigned num) override { return traits<BASE>::reg_bit_widths[num]; }
 
-        std::unordered_map<std::string, uint64_t> get_symbol_table(std::string name) override { return arch.get_sym_table(name); }
+        std::unordered_map<std::string, uint64_t> const& get_symbol_table(std::string name) override { return arch.symbol_table; }
 
         riscv_hart_mu_p<BASE, FEAT, LOGCAT>& arch;
     };
@@ -368,9 +375,6 @@ protected:
     int64_t instret_offset{0};
     uint64_t minstret_csr{0};
     reg_t fault_data;
-    uint64_t tohost = tohost_dflt;
-    uint64_t fromhost = fromhost_dflt;
-    bool tohost_lower_written = false;
     riscv_instrumentation_if instr_if;
 
     semihosting_cb_t<reg_t> semihosting_cb;
@@ -380,7 +384,6 @@ protected:
     using csr_page_type = typename csr_type::page_type;
     mem_type mem;
     csr_type csr;
-    std::stringstream uart_buf;
     std::unordered_map<reg_t, uint64_t> ptw;
     std::unordered_map<uint64_t, uint8_t> atomic_reservation;
     std::unordered_map<unsigned, rd_csr_f> csr_rd_cb;
@@ -402,8 +405,8 @@ protected:
 
     std::vector<uint8_t> tcm;
 
-    iss::status read_csr_reg(unsigned addr, reg_t& val);
-    iss::status write_csr_reg(unsigned addr, reg_t val);
+    iss::status read_plain(unsigned addr, reg_t& val);
+    iss::status write_plain(unsigned addr, reg_t val);
     iss::status read_null(unsigned addr, reg_t& val);
     iss::status write_null(unsigned addr, reg_t val) { return iss::status::Ok; }
     iss::status read_cycle(unsigned addr, reg_t& val);
@@ -426,15 +429,17 @@ protected:
     iss::status read_intstatus(unsigned addr, reg_t& val);
     iss::status write_intthresh(unsigned addr, reg_t val);
     iss::status write_xtvt(unsigned addr, reg_t val);
-    iss::status write_dcsr_dcsr(unsigned addr, reg_t val);
-    iss::status read_dcsr_reg(unsigned addr, reg_t& val);
-    iss::status write_dcsr_reg(unsigned addr, reg_t val);
-    iss::status read_dpc_reg(unsigned addr, reg_t& val);
-    iss::status write_dpc_reg(unsigned addr, reg_t val);
-    iss::status write_pmpcfg_reg(unsigned addr, reg_t val);
+    iss::status write_dcsr(unsigned addr, reg_t val);
+    iss::status read_debug(unsigned addr, reg_t& val);
+    iss::status write_dscratch(unsigned addr, reg_t val);
+    iss::status read_dpc(unsigned addr, reg_t& val);
+    iss::status write_dpc(unsigned addr, reg_t val);
+    iss::status read_fcsr(unsigned addr, reg_t& val);
+    iss::status write_fcsr(unsigned addr, reg_t val);
+    iss::status write_pmpcfg(unsigned addr, reg_t val);
 
-    virtual iss::status read_custom_csr_reg(unsigned addr, reg_t& val) { return iss::status::Err; };
-    virtual iss::status write_custom_csr_reg(unsigned addr, reg_t val) { return iss::status::Err; };
+    virtual iss::status read_custom_csr(unsigned addr, reg_t& val) { return iss::status::Err; };
+    virtual iss::status write_custom_csr(unsigned addr, reg_t val) { return iss::status::Err; };
 
     void register_custom_csr_rd(unsigned addr) { csr_rd_cb[addr] = &this_class::read_custom_csr_reg; }
     void register_custom_csr_wr(unsigned addr) { csr_wr_cb[addr] = &this_class::write_custom_csr_reg; }
@@ -473,19 +478,22 @@ riscv_hart_mu_p<BASE, FEAT, LOGCAT>::riscv_hart_mu_p(feature_config cfg)
     csr[marchid] = traits<BASE>::MARCHID_VAL;
     csr[mimpid] = 1;
 
-    uart_buf.str("");
+    if(traits<BASE>::FLEN > 0) {
+        csr_rd_cb[fcsr] = &this_class::read_fcsr;
+        csr_wr_cb[fcsr] = &this_class::write_fcsr;
+    }
     for(unsigned addr = mhpmcounter3; addr <= mhpmcounter31; ++addr) {
         csr_rd_cb[addr] = &this_class::read_null;
-        csr_wr_cb[addr] = &this_class::write_csr_reg;
+        csr_wr_cb[addr] = &this_class::write_plain;
     }
     if(traits<BASE>::XLEN == 32)
         for(unsigned addr = mhpmcounter3h; addr <= mhpmcounter31h; ++addr) {
             csr_rd_cb[addr] = &this_class::read_null;
-            csr_wr_cb[addr] = &this_class::write_csr_reg;
+            csr_wr_cb[addr] = &this_class::write_plain;
         }
     for(unsigned addr = mhpmevent3; addr <= mhpmevent31; ++addr) {
         csr_rd_cb[addr] = &this_class::read_null;
-        csr_wr_cb[addr] = &this_class::write_csr_reg;
+        csr_wr_cb[addr] = &this_class::write_plain;
     }
     for(unsigned addr = hpmcounter3; addr <= hpmcounter31; ++addr) {
         csr_rd_cb[addr] = &this_class::read_null;
@@ -493,12 +501,11 @@ riscv_hart_mu_p<BASE, FEAT, LOGCAT>::riscv_hart_mu_p(feature_config cfg)
     if(traits<BASE>::XLEN == 32)
         for(unsigned addr = hpmcounter3h; addr <= hpmcounter31h; ++addr) {
             csr_rd_cb[addr] = &this_class::read_null;
-            // csr_wr_cb[addr] = &this_class::write_csr_reg;
         }
     // common regs
     const std::array<unsigned, 4> roaddrs{{misa, mvendorid, marchid, mimpid}};
     for(auto addr : roaddrs) {
-        csr_rd_cb[addr] = &this_class::read_csr_reg;
+        csr_rd_cb[addr] = &this_class::read_plain;
         csr_wr_cb[addr] = &this_class::write_null;
     }
     const std::array<unsigned, 8> rwaddrs{{
@@ -512,8 +519,8 @@ riscv_hart_mu_p<BASE, FEAT, LOGCAT>::riscv_hart_mu_p(feature_config cfg)
         utval,
     }};
     for(auto addr : rwaddrs) {
-        csr_rd_cb[addr] = &this_class::read_csr_reg;
-        csr_wr_cb[addr] = &this_class::write_csr_reg;
+        csr_rd_cb[addr] = &this_class::read_plain;
+        csr_wr_cb[addr] = &this_class::write_plain;
     }
     // special handling & overrides
     csr_rd_cb[time] = &this_class::read_time;
@@ -558,18 +565,18 @@ riscv_hart_mu_p<BASE, FEAT, LOGCAT>::riscv_hart_mu_p(feature_config cfg)
 
     if(FEAT & FEAT_PMP) {
         for(size_t i = pmpaddr0; i <= pmpaddr15; ++i) {
-            csr_rd_cb[i] = &this_class::read_csr_reg;
-            csr_wr_cb[i] = &this_class::write_csr_reg;
+            csr_rd_cb[i] = &this_class::read_plain;
+            csr_wr_cb[i] = &this_class::write_plain;
         }
         for(size_t i = pmpcfg0; i < pmpcfg0 + 16 / sizeof(reg_t); ++i) {
-            csr_rd_cb[i] = &this_class::read_csr_reg;
-            csr_wr_cb[i] = &this_class::write_pmpcfg_reg;
+            csr_rd_cb[i] = &this_class::read_plain;
+            csr_wr_cb[i] = &this_class::write_pmpcfg;
         }
     }
     if(FEAT & FEAT_EXT_N) {
-        csr_rd_cb[mideleg] = &this_class::read_csr_reg;
+        csr_rd_cb[mideleg] = &this_class::read_plain;
         csr_wr_cb[mideleg] = &this_class::write_ideleg;
-        csr_rd_cb[medeleg] = &this_class::read_csr_reg;
+        csr_rd_cb[medeleg] = &this_class::read_plain;
         csr_wr_cb[medeleg] = &this_class::write_edeleg;
         csr_rd_cb[uie] = &this_class::read_ie;
         csr_wr_cb[uie] = &this_class::write_ie;
@@ -583,7 +590,7 @@ riscv_hart_mu_p<BASE, FEAT, LOGCAT>::riscv_hart_mu_p(feature_config cfg)
         csr_rd_cb[utvec] = &this_class::read_tvec;
     }
     if(FEAT & FEAT_CLIC) {
-        csr_rd_cb[mtvt] = &this_class::read_csr_reg;
+        csr_rd_cb[mtvt] = &this_class::read_plain;
         csr_wr_cb[mtvt] = &this_class::write_xtvt;
         //        csr_rd_cb[mxnti] = &this_class::read_csr_reg;
         //        csr_wr_cb[mxnti] = &this_class::write_csr_reg;
@@ -593,14 +600,14 @@ riscv_hart_mu_p<BASE, FEAT, LOGCAT>::riscv_hart_mu_p(feature_config cfg)
         //        csr_wr_cb[mscratchcsw] = &this_class::write_csr_reg;
         //        csr_rd_cb[mscratchcswl] = &this_class::read_csr_reg;
         //        csr_wr_cb[mscratchcswl] = &this_class::write_csr_reg;
-        csr_rd_cb[mintthresh] = &this_class::read_csr_reg;
+        csr_rd_cb[mintthresh] = &this_class::read_plain;
         csr_wr_cb[mintthresh] = &this_class::write_intthresh;
         if(FEAT & FEAT_EXT_N) {
-            csr_rd_cb[utvt] = &this_class::read_csr_reg;
+            csr_rd_cb[utvt] = &this_class::read_plain;
             csr_wr_cb[utvt] = &this_class::write_xtvt;
             csr_rd_cb[uintstatus] = &this_class::read_intstatus;
             csr_wr_cb[uintstatus] = &this_class::write_null;
-            csr_rd_cb[uintthresh] = &this_class::read_csr_reg;
+            csr_rd_cb[uintthresh] = &this_class::read_plain;
             csr_wr_cb[uintthresh] = &this_class::write_intthresh;
         }
         clic_int_reg.resize(cfg.clic_num_irq, clic_int_reg_t{.raw = 0});
@@ -629,14 +636,14 @@ riscv_hart_mu_p<BASE, FEAT, LOGCAT>::riscv_hart_mu_p(feature_config cfg)
         insert_mem_range(cfg.tcm_base, cfg.tcm_size, read_clic_cb, write_clic_cb);
     }
     if(FEAT & FEAT_DEBUG) {
-        csr_wr_cb[dscratch0] = &this_class::write_dcsr_reg;
-        csr_rd_cb[dscratch0] = &this_class::read_dcsr_reg;
-        csr_wr_cb[dscratch1] = &this_class::write_dcsr_reg;
-        csr_rd_cb[dscratch1] = &this_class::read_dcsr_reg;
-        csr_wr_cb[dpc] = &this_class::write_dpc_reg;
-        csr_rd_cb[dpc] = &this_class::read_dpc_reg;
-        csr_wr_cb[dcsr] = &this_class::write_dcsr_dcsr;
-        csr_rd_cb[dcsr] = &this_class::read_dcsr_reg;
+        csr_wr_cb[dscratch0] = &this_class::write_dscratch;
+        csr_rd_cb[dscratch0] = &this_class::read_debug;
+        csr_wr_cb[dscratch1] = &this_class::write_dscratch;
+        csr_rd_cb[dscratch1] = &this_class::read_debug;
+        csr_wr_cb[dpc] = &this_class::write_dpc;
+        csr_rd_cb[dpc] = &this_class::read_dpc;
+        csr_wr_cb[dcsr] = &this_class::write_dcsr;
+        csr_rd_cb[dcsr] = &this_class::read_debug;
     }
     hart_mem_rd_delegate = [this](phys_addr_t a, unsigned l, uint8_t* const d) -> iss::status { return this->read_mem(a, l, d); };
     hart_mem_wr_delegate = [this](phys_addr_t a, unsigned l, uint8_t const* const d) -> iss::status { return this->write_mem(a, l, d); };
@@ -644,71 +651,14 @@ riscv_hart_mu_p<BASE, FEAT, LOGCAT>::riscv_hart_mu_p(feature_config cfg)
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
 std::pair<uint64_t, bool> riscv_hart_mu_p<BASE, FEAT, LOGCAT>::load_file(std::string name, int type) {
-    FILE* fp = fopen(name.c_str(), "r");
-    if(fp) {
-        std::array<char, 5> buf;
-        auto n = fread(buf.data(), 1, 4, fp);
-        fclose(fp);
-        if(n != 4)
-            throw std::runtime_error("input file has insufficient size");
-        buf[4] = 0;
-        if(strcmp(buf.data() + 1, "ELF") == 0) {
-            // Create elfio reader
-            ELFIO::elfio reader;
-            // Load ELF data
-            if(!reader.load(name))
-                throw std::runtime_error("could not process elf file");
-            // check elf properties
-            if(reader.get_class() != ELFCLASS32)
-                if(sizeof(reg_t) == 4)
-                    throw std::runtime_error("wrong elf class in file");
-            if(reader.get_type() != ET_EXEC)
-                throw std::runtime_error("wrong elf type in file");
-            if(reader.get_machine() != EM_RISCV)
-                throw std::runtime_error("wrong elf machine in file");
-            auto entry = reader.get_entry();
-            for(const auto pseg : reader.segments) {
-                const auto fsize = pseg->get_file_size(); // 0x42c/0x0
-                const auto seg_data = pseg->get_data();
-                const auto type = pseg->get_type();
-                if(type == 1 && fsize > 0) {
-                    auto res = this->write(iss::address_type::PHYSICAL, iss::access_type::DEBUG_WRITE, traits<BASE>::MEM,
-                                           pseg->get_physical_address(), fsize, reinterpret_cast<const uint8_t* const>(seg_data));
-                    if(res != iss::Ok)
-                        CPPLOG(ERR) << "problem writing " << fsize << "bytes to 0x" << std::hex << pseg->get_physical_address();
-                }
-            }
-            for(const auto sec : reader.sections) {
-                if(sec->get_name() == ".symtab") {
-                    if(SHT_SYMTAB == sec->get_type() || SHT_DYNSYM == sec->get_type()) {
-                        ELFIO::symbol_section_accessor symbols(reader, sec);
-                        auto sym_no = symbols.get_symbols_num();
-                        std::string name;
-                        ELFIO::Elf64_Addr value = 0;
-                        ELFIO::Elf_Xword size = 0;
-                        unsigned char bind = 0;
-                        unsigned char type = 0;
-                        ELFIO::Elf_Half section = 0;
-                        unsigned char other = 0;
-                        for(auto i = 0U; i < sym_no; ++i) {
-                            symbols.get_symbol(i, name, value, size, bind, type, section, other);
-                            if(name == "tohost") {
-                                tohost = value;
-                            } else if(name == "fromhost") {
-                                fromhost = value;
-                            }
-                        }
-                    }
-                } else if(sec->get_name() == ".tohost") {
-                    tohost = sec->get_address();
-                    fromhost = tohost + 0x40;
-                }
-            }
-            return std::make_pair(entry, true);
-        }
-        throw std::runtime_error(fmt::format("memory load file {} is not a valid elf file", name));
+    if(read_elf_file(name, sizeof(reg_t) == 4 ? ELFIO::ELFCLASS32 : ELFIO::ELFCLASS64,
+                     [this](uint64_t addr, uint64_t size, const uint8_t* const data) -> iss::status {
+                         return this->write(iss::address_type::PHYSICAL, iss::access_type::DEBUG_WRITE, traits<BASE>::MEM, addr, size,
+                                            data);
+                     })) {
+        return std::make_pair(entry_address, true);
     }
-    throw std::runtime_error(fmt::format("memory load file not found, check if {} is a valid file", name));
+    return std::make_pair(entry_address, false);
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
@@ -725,7 +675,7 @@ inline void riscv_hart_mu_p<BASE, FEAT, LOGCAT>::insert_mem_range(uint64_t base,
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-inline iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_pmpcfg_reg(unsigned addr, reg_t val) {
+inline iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_pmpcfg(unsigned addr, reg_t val) {
     csr[addr] = val & 0x9f9f9f9f;
     return iss::Ok;
 }
@@ -835,7 +785,7 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read(const address_type type, c
                     return iss::Err;
                 }
             }
-            auto alignment = is_fetch(access) ? (has_compressed() ? 2 : 4) : length;
+            auto alignment = is_fetch(access) ? (has_compressed() ? 2 : 4) : std::min<unsigned>(length, sizeof(reg_t));
             if(unlikely(is_fetch(access) && (addr & (alignment - 1)))) {
                 fault_data = addr;
                 if(is_debug(access))
@@ -870,8 +820,10 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read(const address_type type, c
                 }
                 return res;
             } catch(trap_access& ta) {
-                this->reg.trap_state = (1UL << 31) | ta.id;
-                fault_data = ta.addr;
+                if((access & access_type::DEBUG) == 0) {
+                    this->reg.trap_state = (1UL << 31) | ta.id;
+                    fault_data = ta.addr;
+                }
                 return iss::Err;
             }
         } break;
@@ -898,8 +850,10 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read(const address_type type, c
         }
         return iss::Ok;
     } catch(trap_access& ta) {
-        this->reg.trap_state = (1UL << 31) | ta.id;
-        fault_data = ta.addr;
+        if((access & access_type::DEBUG) == 0) {
+            this->reg.trap_state = (1UL << 31) | ta.id;
+            fault_data = ta.addr;
+        }
         return iss::Err;
     }
 }
@@ -950,7 +904,8 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write(const address_type type,
                 return iss::Err;
             }
             try {
-                if(length > 1 && (addr & (length - 1)) && (access & access_type::DEBUG) != access_type::DEBUG) {
+                auto alignment = std::min<unsigned>(length, sizeof(reg_t));
+                if(length > 1 && (addr & (alignment - 1)) && !is_debug(access)) {
                     this->reg.trap_state = (1UL << 31) | 6 << 16;
                     fault_data = addr;
                     return iss::Err;
@@ -980,40 +935,6 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write(const address_type type,
                 fault_data = ta.addr;
                 return iss::Err;
             }
-
-            if((addr + length) > mem.size())
-                return iss::Err;
-            switch(addr) {
-            case 0x10013000: // UART0 base, TXFIFO reg
-            case 0x10023000: // UART1 base, TXFIFO reg
-                uart_buf << (char)data[0];
-                if(((char)data[0]) == '\n' || data[0] == 0) {
-                    // CPPLOG(INFO)<<"UART"<<((addr>>16)&0x3)<<" send
-                    // '"<<uart_buf.str()<<"'";
-                    std::cout << uart_buf.str();
-                    uart_buf.str("");
-                }
-                return iss::Ok;
-            case 0x10008000: { // HFROSC base, hfrosccfg reg
-                auto& p = mem(addr / mem.page_size);
-                auto offs = addr & mem.page_addr_mask;
-                std::copy(data, data + length, p.data() + offs);
-                auto& x = *(p.data() + offs + 3);
-                if(x & 0x40)
-                    x |= 0x80; // hfroscrdy = 1 if hfroscen==1
-                return iss::Ok;
-            }
-            case 0x10008008: { // HFROSC base, pllcfg reg
-                auto& p = mem(addr / mem.page_size);
-                auto offs = addr & mem.page_addr_mask;
-                std::copy(data, data + length, p.data() + offs);
-                auto& x = *(p.data() + offs + 3);
-                x |= 0x80; // set pll lock upon writing
-                return iss::Ok;
-            } break;
-            default: {
-            }
-            }
         } break;
         case traits<BASE>::CSR: {
             if(length != sizeof(reg_t))
@@ -1040,8 +961,10 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write(const address_type type,
         }
         return iss::Ok;
     } catch(trap_access& ta) {
-        this->reg.trap_state = (1UL << 31) | ta.id;
-        fault_data = ta.addr;
+        if((access & access_type::DEBUG) == 0) {
+            this->reg.trap_state = (1UL << 31) | ta.id;
+            fault_data = ta.addr;
+        }
         return iss::Err;
     }
 }
@@ -1074,12 +997,6 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_csr(unsigned addr, reg_t
     return (this->*(it->second))(addr, val);
 }
 
-template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_csr_reg(unsigned addr, reg_t& val) {
-    val = csr[addr];
-    return iss::Ok;
-}
-
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_null(unsigned addr, reg_t& val) {
     val = 0;
@@ -1087,14 +1004,20 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_null(unsigned addr, reg_t&
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_csr_reg(unsigned addr, reg_t val) {
+iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_plain(unsigned addr, reg_t& val) {
+    val = csr[addr];
+    return iss::Ok;
+}
+
+template <typename BASE, features_e FEAT, typename LOGCAT>
+iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_plain(unsigned addr, reg_t val) {
     csr[addr] = val;
     return iss::Ok;
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_cycle(unsigned addr, reg_t& val) {
-    auto cycle_val = this->reg.icount + cycle_offset;
+    auto cycle_val = this->reg.cycle + cycle_offset;
     if(addr == mcycle) {
         val = static_cast<reg_t>(cycle_val);
     } else if(addr == mcycleh) {
@@ -1114,7 +1037,7 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_cycle(unsigned addr, reg_
             mcycle_csr = (static_cast<uint64_t>(val) << 32) + (mcycle_csr & 0xffffffff);
         }
     }
-    cycle_offset = mcycle_csr - this->reg.icount; // TODO: relying on wrap-around
+    cycle_offset = mcycle_csr - this->reg.cycle; // TODO: relying on wrap-around
     return iss::Ok;
 }
 
@@ -1145,7 +1068,7 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_instret(unsigned addr, re
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_time(unsigned addr, reg_t& val) {
-    uint64_t time_val = this->reg.icount / (100000000 / 32768 - 1); //-> ~3052;
+    uint64_t time_val = this->reg.cycle / (100000000 / 32768 - 1); //-> ~3052;
     if(addr == time) {
         val = static_cast<reg_t>(time_val);
     } else if(addr == timeh) {
@@ -1161,6 +1084,7 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_tvec(unsigned addr, reg_t&
     val = FEAT & features_e::FEAT_CLIC ? csr[addr] : csr[addr] & ~2;
     return iss::Ok;
 }
+
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_status(unsigned addr, reg_t& val) {
     val = state.mstatus & hart_state_type::get_mask((addr >> 8) & 0x3);
@@ -1272,7 +1196,7 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_epc(unsigned addr, reg_t
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_dcsr_dcsr(unsigned addr, reg_t val) {
+iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_dcsr(unsigned addr, reg_t val) {
     if(!debug_mode_active())
         throw illegal_instruction_fault(this->fault_data);
     //                  +-------------- ebreakm
@@ -1284,7 +1208,7 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_dcsr_dcsr(unsigned addr,
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_dcsr_reg(unsigned addr, reg_t& val) {
+iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_debug(unsigned addr, reg_t& val) {
     if(!debug_mode_active())
         throw illegal_instruction_fault(this->fault_data);
     val = csr[addr];
@@ -1292,7 +1216,7 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_dcsr_reg(unsigned addr, re
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_dcsr_reg(unsigned addr, reg_t val) {
+iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_dscratch(unsigned addr, reg_t val) {
     if(!debug_mode_active())
         throw illegal_instruction_fault(this->fault_data);
     csr[addr] = val;
@@ -1300,7 +1224,7 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_dcsr_reg(unsigned addr, r
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_dpc_reg(unsigned addr, reg_t& val) {
+iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_dpc(unsigned addr, reg_t& val) {
     if(!debug_mode_active())
         throw illegal_instruction_fault(this->fault_data);
     val = this->reg.DPC;
@@ -1308,7 +1232,7 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_dpc_reg(unsigned addr, reg
 }
 
 template <typename BASE, features_e FEAT, typename LOGCAT>
-iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_dpc_reg(unsigned addr, reg_t val) {
+iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_dpc(unsigned addr, reg_t val) {
     if(!debug_mode_active())
         throw illegal_instruction_fault(this->fault_data);
     this->reg.DPC = val;
@@ -1324,6 +1248,18 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_intstatus(unsigned addr, r
     return iss::Ok;
 }
 
+template <typename BASE, features_e FEAT, typename LOGCAT>
+iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_fcsr(unsigned addr, reg_t& val) {
+    val = this->get_fcsr();
+    return iss::Ok;
+}
+
+template <typename BASE, features_e FEAT, typename LOGCAT>
+iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_fcsr(unsigned addr, reg_t val) {
+    this->set_fcsr(val);
+    return iss::Ok;
+}
+
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_intthresh(unsigned addr, reg_t val) {
     csr[addr] = (val & 0xff) | (1 << (cfg.clic_int_ctl_bits)) - 1;
@@ -1347,65 +1283,53 @@ iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::read_mem(phys_addr_t paddr, uns
     }
     return iss::Ok;
 }
-
 template <typename BASE, features_e FEAT, typename LOGCAT>
 iss::status riscv_hart_mu_p<BASE, FEAT, LOGCAT>::write_mem(phys_addr_t paddr, unsigned length, const uint8_t* const data) {
-    switch(paddr.val) {
-    // TODO remove UART, Peripherals should not be part of the ISS
-    case 0xFFFF0000: // UART0 base, TXFIFO reg
-        if(((char)data[0]) == '\n' || data[0] == 0) {
-            CPPLOG(INFO) << "UART" << ((paddr.val >> 12) & 0x3) << " send '" << uart_buf.str() << "'";
-            uart_buf.str("");
-        } else if(((char)data[0]) != '\r')
-            uart_buf << (char)data[0];
-        break;
-    default: {
-        mem_type::page_type& p = mem(paddr.val / mem.page_size);
-        std::copy(data, data + length, p.data() + (paddr.val & mem.page_addr_mask));
-        // tohost handling in case of riscv-test
-        if(paddr.access && iss::access_type::FUNC) {
-            auto tohost_upper =
-                (traits<BASE>::XLEN == 32 && paddr.val == (tohost + 4)) || (traits<BASE>::XLEN == 64 && paddr.val == tohost);
-            auto tohost_lower = (traits<BASE>::XLEN == 32 && paddr.val == tohost) || (traits<BASE>::XLEN == 64 && paddr.val == tohost);
-            if(tohost_lower || tohost_upper) {
-                uint64_t hostvar = *reinterpret_cast<uint64_t*>(p.data() + (tohost & mem.page_addr_mask));
-                // in case of 32 bit system, two writes to tohost are needed, only evaluate on the second (high) write
-                if(tohost_upper && (tohost_lower || tohost_lower_written)) {
-                    switch(hostvar >> 48) {
-                    case 0:
-                        if(hostvar != 0x1) {
-                            CPPLOG(FATAL) << "tohost value is 0x" << std::hex << hostvar << std::dec << " (" << hostvar
-                                          << "), stopping simulation";
-                        } else {
-                            CPPLOG(INFO) << "tohost value is 0x" << std::hex << hostvar << std::dec << " (" << hostvar
-                                         << "), stopping simulation";
-                        }
-                        this->reg.trap_state = std::numeric_limits<uint32_t>::max();
-                        this->interrupt_sim = hostvar;
-#ifndef WITH_TCC
-                        throw(iss::simulation_stopped(hostvar));
-#endif
-                        break;
-                    case 0x0101: {
-                        char c = static_cast<char>(hostvar & 0xff);
-                        if(c == '\n' || c == 0) {
-                            CPPLOG(INFO) << "tohost send '" << uart_buf.str() << "'";
-                            uart_buf.str("");
-                        } else
-                            uart_buf << c;
-                    } break;
-                    default:
-                        break;
-                    }
-                    tohost_lower_written = false;
-                } else if(tohost_lower)
-                    tohost_lower_written = true;
-            } else if((traits<BASE>::XLEN == 32 && paddr.val == fromhost + 4) || (traits<BASE>::XLEN == 64 && paddr.val == fromhost)) {
-                uint64_t fhostvar = *reinterpret_cast<uint64_t*>(p.data() + (fromhost & mem.page_addr_mask));
-                *reinterpret_cast<uint64_t*>(p.data() + (tohost & mem.page_addr_mask)) = fhostvar;
+    mem_type::page_type& p = mem(paddr.val / mem.page_size);
+    std::copy(data, data + length, p.data() + (paddr.val & mem.page_addr_mask));
+    // tohost handling in case of riscv-test
+    // according to https://github.com/riscv-software-src/riscv-isa-sim/issues/364#issuecomment-607657754:
+    if(paddr.access && iss::access_type::FUNC) {
+        if(paddr.val == tohost) {
+            reg_t cur_data = *reinterpret_cast<const reg_t*>(data);
+            // Extract Device (bits 63:56)
+            uint8_t device = traits<BASE>::XLEN == 32 ? 0 : (cur_data >> 56) & 0xFF;
+            // Extract Command (bits 55:48)
+            uint8_t command = traits<BASE>::XLEN == 32 ? 0 : (cur_data >> 48) & 0xFF;
+            // Extract payload (bits 47:0)
+            uint64_t payload_addr = cur_data & 0xFFFFFFFFFFFFULL;
+            if(payload_addr & 1) {
+                CPPLOG(FATAL) << "tohost value is 0x" << std::hex << payload_addr << std::dec << " (" << payload_addr
+                              << "), stopping simulation";
+                this->reg.trap_state = std::numeric_limits<uint32_t>::max();
+                this->interrupt_sim = payload_addr;
+                return iss::Ok;
+            } else if(device == 0 && command == 0) {
+                std::array<uint64_t, 8> loaded_payload;
+                if(read(address_type::PHYSICAL, access_type::DEBUG_READ, traits<BASE>::MEM, payload_addr, 8 * sizeof(uint64_t),
+                        reinterpret_cast<uint8_t*>(loaded_payload.data())) == iss::Err)
+                    CPPLOG(ERR) << "Syscall read went wrong";
+                uint64_t syscall_num = loaded_payload.at(0);
+                if(syscall_num == 64) { // SYS_WRITE
+                    return execute_sys_write(this, loaded_payload, traits<BASE>::MEM);
+                } else {
+                    CPPLOG(ERR) << "tohost syscall with number 0x" << std::hex << syscall_num << std::dec << " (" << syscall_num
+                                << ") not implemented";
+                    this->reg.trap_state = std::numeric_limits<uint32_t>::max();
+                    this->interrupt_sim = payload_addr;
+                    return iss::Ok;
+                }
+            } else {
+                CPPLOG(ERR) << "tohost functionality not implemented for device " << device << " and command " << command;
+                this->reg.trap_state = std::numeric_limits<uint32_t>::max();
+                this->interrupt_sim = payload_addr;
+                return iss::Ok;
             }
         }
-    }
+        if((traits<BASE>::XLEN == 32 && paddr.val == fromhost + 4) || (traits<BASE>::XLEN == 64 && paddr.val == fromhost)) {
+            uint64_t fhostvar = *reinterpret_cast<uint64_t*>(p.data() + (fromhost & mem.page_addr_mask));
+            *reinterpret_cast<uint64_t*>(p.data() + (tohost & mem.page_addr_mask)) = fhostvar;
+        }
     }
     return iss::Ok;
 }

src/iss/arch/tgc5c.cpp

@@ -1,5 +1,5 @@
 /*******************************************************************************
- * Copyright (C) 2017 - 2020 MINRES Technologies GmbH
+ * Copyright (C) 2024 MINRES Technologies GmbH
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without

src/iss/debugger/riscv_target_adapter.h

@@ -30,8 +30,8 @@
  *
  *******************************************************************************/
 
-#ifndef _ISS_DEBUGGER_RISCV_TARGET_ADAPTER_H_
-#define _ISS_DEBUGGER_RISCV_TARGET_ADAPTER_H_
+#ifndef _ISS_ARCH_DEBUGGER_RISCV_TARGET_ADAPTER_H_
+#define _ISS_ARCH_DEBUGGER_RISCV_TARGET_ADAPTER_H_
 
 #include "iss/arch_if.h"
 #include <iss/arch/traits.h>
@@ -48,6 +48,10 @@
 
 namespace iss {
 namespace debugger {
+
+char const* const get_csr_name(unsigned);
+constexpr auto csr_offset = 100U;
+
 using namespace iss::arch;
 using namespace iss::debugger;
 
@@ -129,11 +133,17 @@ public:
 
 protected:
     static inline constexpr addr_t map_addr(const addr_t& i) { return i; }
-
+    std::string csr_xml;
     iss::arch_if* core;
     rp_thread_ref thread_idx;
 };
 
+template <typename ARCH> typename std::enable_if<iss::arch::traits<ARCH>::FLEN != 0, unsigned>::type get_f0_offset() {
+    return iss::arch::traits<ARCH>::F0;
+}
+
+template <typename ARCH> typename std::enable_if<iss::arch::traits<ARCH>::FLEN == 0, unsigned>::type get_f0_offset() { return 0; }
+
 template <typename ARCH> status riscv_target_adapter<ARCH>::set_gen_thread(rp_thread_ref& thread) {
     thread_idx = thread;
     return Ok;
@@ -175,34 +185,37 @@ template <typename ARCH> status riscv_target_adapter<ARCH>::current_thread_query
 
 template <typename ARCH> status riscv_target_adapter<ARCH>::read_registers(std::vector<uint8_t>& data, std::vector<uint8_t>& avail) {
     CPPLOG(TRACE) << "reading target registers";
-    // return idx<0?:;
     data.clear();
     avail.clear();
     const uint8_t* reg_base = core->get_regs_base_ptr();
     auto start_reg = arch::traits<ARCH>::X0;
-    for(size_t reg_no = start_reg; reg_no < start_reg + 33 /*arch::traits<ARCH>::NUM_REGS*/; ++reg_no) {
-        auto reg_width = arch::traits<ARCH>::reg_bit_widths[reg_no] / 8;
-        unsigned offset = traits<ARCH>::reg_byte_offsets[reg_no];
-        for(size_t j = 0; j < reg_width; ++j) {
-            data.push_back(*(reg_base + offset + j));
-            avail.push_back(0xff);
+    for(size_t i = 0; i < 33; ++i) {
+        if(i < arch::traits<ARCH>::RFS || i == arch::traits<ARCH>::PC) {
+            auto reg_no = i < 32 ? start_reg + i : arch::traits<ARCH>::PC;
+            unsigned offset = traits<ARCH>::reg_byte_offsets[reg_no];
+            for(size_t j = 0; j < arch::traits<ARCH>::XLEN / 8; ++j) {
+                data.push_back(*(reg_base + offset + j));
+                avail.push_back(0xff);
+            }
+        } else {
+            for(size_t j = 0; j < arch::traits<ARCH>::XLEN / 8; ++j) {
+                data.push_back(0);
+                avail.push_back(0);
+            }
+        }
+    }
+    if(iss::arch::traits<ARCH>::FLEN > 0) {
+        auto fstart_reg = get_f0_offset<ARCH>();
+        for(size_t i = 0; i < 32; ++i) {
+            auto reg_no = fstart_reg + i;
+            auto reg_width = arch::traits<ARCH>::reg_bit_widths[reg_no] / 8;
+            unsigned offset = traits<ARCH>::reg_byte_offsets[reg_no];
+            for(size_t j = 0; j < reg_width; ++j) {
+                data.push_back(*(reg_base + offset + j));
+                avail.push_back(0xff);
+            }
         }
     }
-    // work around fill with F type registers
-    //    if (arch::traits<ARCH>::NUM_REGS < 65) {
-    //        auto reg_width = sizeof(typename arch::traits<ARCH>::reg_t);
-    //        for (size_t reg_no = 0; reg_no < 33; ++reg_no) {
-    //            for (size_t j = 0; j < reg_width; ++j) {
-    //                data.push_back(0x0);
-    //                avail.push_back(0x00);
-    //            }
-    //            // if(arch::traits<ARCH>::XLEN < 64)
-    //            //     for(unsigned j=0; j<4; ++j){
-    //            //         data.push_back(0x0);
-    //            //         avail.push_back(0x00);
-    //            //     }
-    //        }
-    //    }
     return Ok;
 }
 
@@ -210,25 +223,25 @@ template <typename ARCH> status riscv_target_adapter<ARCH>::write_registers(cons
     auto start_reg = arch::traits<ARCH>::X0;
     auto* reg_base = core->get_regs_base_ptr();
     auto iter = data.data();
-    bool e_ext = arch::traits<ARCH>::PC < 32;
-    for(size_t reg_no = 0; reg_no < start_reg + 33 /*arch::traits<ARCH>::NUM_REGS*/; ++reg_no) {
-        if(e_ext && reg_no > 15) {
-            if(reg_no == 32) {
-                auto reg_width = arch::traits<ARCH>::reg_bit_widths[arch::traits<ARCH>::PC] / 8;
-                auto offset = traits<ARCH>::reg_byte_offsets[arch::traits<ARCH>::PC];
-                std::copy(iter, iter + reg_width, reg_base);
-            } else {
-                const uint64_t zero_val = 0;
-                auto reg_width = arch::traits<ARCH>::reg_bit_widths[15] / 8;
-                auto iter = (uint8_t*)&zero_val;
-                std::copy(iter, iter + reg_width, reg_base);
-            }
-        } else {
-            auto reg_width = arch::traits<ARCH>::reg_bit_widths[reg_no] / 8;
-            auto offset = traits<ARCH>::reg_byte_offsets[reg_no];
-            std::copy(iter, iter + reg_width, reg_base);
-            iter += 4;
-            reg_base += offset;
+    auto iter_end = data.data() + data.size();
+    for(size_t i = 0; i < 33 && iter < iter_end; ++i) {
+        auto reg_width = arch::traits<ARCH>::XLEN / 8;
+        if(i < arch::traits<ARCH>::RFS) {
+            auto offset = traits<ARCH>::reg_byte_offsets[start_reg + i];
+            std::copy(iter, iter + reg_width, reg_base + offset);
+        } else if(i == 32) {
+            auto offset = traits<ARCH>::reg_byte_offsets[arch::traits<ARCH>::PC];
+            std::copy(iter, iter + reg_width, reg_base + offset);
+        }
+        iter += reg_width;
+    }
+    if(iss::arch::traits<ARCH>::FLEN > 0) {
+        auto fstart_reg = get_f0_offset<ARCH>();
+        auto reg_width = arch::traits<ARCH>::FLEN / 8;
+        for(size_t i = 0; i < 32 && iter < iter_end; ++i) {
+            unsigned offset = traits<ARCH>::reg_byte_offsets[fstart_reg + i];
+            std::copy(iter, iter + reg_width, reg_base + offset);
+            iter += reg_width;
         }
     }
     return Ok;
@@ -236,7 +249,7 @@ template <typename ARCH> status riscv_target_adapter<ARCH>::write_registers(cons
 
 template <typename ARCH>
 status riscv_target_adapter<ARCH>::read_single_register(unsigned int reg_no, std::vector<uint8_t>& data, std::vector<uint8_t>& avail) {
-    if(reg_no < 65) {
+    if(reg_no < csr_offset) {
         // auto reg_size = arch::traits<ARCH>::reg_bit_width(static_cast<typename
         // arch::traits<ARCH>::reg_e>(reg_no))/8;
         auto* reg_base = core->get_regs_base_ptr();
@@ -247,23 +260,24 @@ status riscv_target_adapter<ARCH>::read_single_register(unsigned int reg_no, std
         std::copy(reg_base + offset, reg_base + offset + reg_width, data.begin());
         std::fill(avail.begin(), avail.end(), 0xff);
     } else {
-        typed_addr_t<iss::address_type::PHYSICAL> a(iss::access_type::DEBUG_READ, traits<ARCH>::CSR, reg_no - 65);
+        typed_addr_t<iss::address_type::PHYSICAL> a(iss::access_type::DEBUG_READ, traits<ARCH>::CSR, reg_no - csr_offset);
         data.resize(sizeof(typename traits<ARCH>::reg_t));
         avail.resize(sizeof(typename traits<ARCH>::reg_t));
         std::fill(avail.begin(), avail.end(), 0xff);
         core->read(a, data.size(), data.data());
+        std::fill(avail.begin(), avail.end(), 0xff);
     }
     return data.size() > 0 ? Ok : Err;
 }
 
 template <typename ARCH> status riscv_target_adapter<ARCH>::write_single_register(unsigned int reg_no, const std::vector<uint8_t>& data) {
-    if(reg_no < 65) {
+    if(reg_no < csr_offset) {
         auto* reg_base = core->get_regs_base_ptr();
         auto reg_width = arch::traits<ARCH>::reg_bit_widths[static_cast<typename arch::traits<ARCH>::reg_e>(reg_no)] / 8;
         auto offset = traits<ARCH>::reg_byte_offsets[reg_no];
         std::copy(data.begin(), data.begin() + reg_width, reg_base + offset);
     } else {
-        typed_addr_t<iss::address_type::PHYSICAL> a(iss::access_type::DEBUG_WRITE, traits<ARCH>::CSR, reg_no - 65);
+        typed_addr_t<iss::address_type::PHYSICAL> a(iss::access_type::DEBUG_WRITE, traits<ARCH>::CSR, reg_no - csr_offset);
         core->write(a, data.size(), data.data());
     }
     return Ok;
@@ -276,7 +290,7 @@ template <typename ARCH> status riscv_target_adapter<ARCH>::read_mem(uint64_t ad
 }
 
 template <typename ARCH> status riscv_target_adapter<ARCH>::write_mem(uint64_t addr, const std::vector<uint8_t>& data) {
-    auto a = map_addr({iss::access_type::DEBUG_READ, iss::address_type::VIRTUAL, 0, addr});
+    auto a = map_addr({iss::access_type::DEBUG_WRITE, iss::address_type::VIRTUAL, 0, addr});
     auto f = [&]() -> status { return core->write(a, data.size(), data.data()); };
     return srv->execute_syncronized(f);
 }
@@ -369,93 +383,57 @@ status riscv_target_adapter<ARCH>::resume_from_addr(bool step, int sig, uint64_t
 }
 
 template <typename ARCH> status riscv_target_adapter<ARCH>::target_xml_query(std::string& out_buf) {
-    const std::string res{"<?xml version=\"1.0\"?><!DOCTYPE target SYSTEM \"gdb-target.dtd\">"
-                          "<target><architecture>riscv:rv32</architecture>"
-                          //"  <feature name=\"org.gnu.gdb.riscv.rv32i\">\n"
-                          //"    <reg name=\"x0\"  bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x1\"  bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x2\"  bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x3\"  bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x4\"  bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x5\"  bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x6\"  bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x7\"  bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x8\"  bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x9\"  bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x10\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x11\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x12\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x13\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x14\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x15\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x16\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x17\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x18\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x19\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x20\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x21\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x22\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x23\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x24\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x25\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x26\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x27\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x28\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x29\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x30\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"    <reg name=\"x31\" bitsize=\"32\" group=\"general\"/>\n"
-                          //"  </feature>\n"
-                          "</target>"};
-    out_buf = res;
+    if(!csr_xml.size()) {
+        std::ostringstream oss;
+        oss << "<?xml version=\"1.0\"?><!DOCTYPE feature SYSTEM \"gdb-target.dtd\"><target version=\"1.0\">\n";
+        if(iss::arch::traits<ARCH>::XLEN == 32)
+            oss << "<architecture>riscv:rv32</architecture>\n";
+        else if(iss::arch::traits<ARCH>::XLEN == 64)
+            oss << "  <architectureriscv:rv64</architecture>\n";
+        oss << "  <feature name=\"org.gnu.gdb.riscv.cpu\">\n";
+        auto reg_base_num = iss::arch::traits<ARCH>::X0;
+        for(auto i = 0U; i < iss::arch::traits<ARCH>::RFS; ++i) {
+            oss << "    <reg name=\"x" << i << "\" bitsize=\"" << iss::arch::traits<ARCH>::reg_bit_widths[reg_base_num + i]
+                << "\" type=\"int\" regnum=\"" << i << "\"/>\n";
+        }
+        oss << "    <reg name=\"pc\" bitsize=\"" << iss::arch::traits<ARCH>::reg_bit_widths[iss::arch::traits<ARCH>::PC]
+            << "\" type=\"code_ptr\" regnum=\"" << 32U << "\"/>\n";
+        oss << "  </feature>\n";
+        if(iss::arch::traits<ARCH>::FLEN > 0) {
+            oss << "  <feature name=\"org.gnu.gdb.riscv.fpu\">\n";
+            auto reg_base_num = get_f0_offset<ARCH>();
+            auto type = iss::arch::traits<ARCH>::FLEN == 32 ? "ieee_single" : "riscv_double";
+            for(auto i = 0U; i < 32; ++i) {
+                oss << "    <reg name=\"f" << i << "\" bitsize=\"" << iss::arch::traits<ARCH>::reg_bit_widths[reg_base_num + i]
+                    << "\" type=\"" << type << "\" regnum=\"" << i + 33 << "\"/>\n";
+            }
+            oss << "    <reg name=\"fcsr\" bitsize=\"" << iss::arch::traits<ARCH>::XLEN << "\" regnum=\"103\" type int/>\n";
+            oss << "    <reg name=\"fflags\" bitsize=\"" << iss::arch::traits<ARCH>::XLEN << "\" regnum=\"101\" type int/>\n";
+            oss << "    <reg name=\"frm\" bitsize=\"" << iss::arch::traits<ARCH>::XLEN << "\" regnum=\"102\" type int/>\n";
+            oss << "  </feature>\n";
+        }
+        oss << "  <feature name=\"org.gnu.gdb.riscv.csr\">\n";
+        std::vector<uint8_t> data;
+        std::vector<uint8_t> avail;
+        data.resize(sizeof(typename traits<ARCH>::reg_t));
+        avail.resize(sizeof(typename traits<ARCH>::reg_t));
+        for(auto i = 0U; i < 4096; ++i) {
+            typed_addr_t<iss::address_type::PHYSICAL> a(iss::access_type::DEBUG_READ, traits<ARCH>::CSR, i);
+            std::fill(avail.begin(), avail.end(), 0xff);
+            auto res = core->read(a, data.size(), data.data());
+            if(res == iss::Ok) {
+                oss << "    <reg name=\"" << get_csr_name(i) << "\" bitsize=\"" << iss::arch::traits<ARCH>::XLEN
+                    << "\"  type=\"int\" regnum=\"" << (i + csr_offset) << "\"/>\n";
+            }
+        }
+        oss << "  </feature>\n";
+        oss << "</target>\n";
+        csr_xml = oss.str();
+    }
+    out_buf = csr_xml;
     return Ok;
 }
-
-/*
- *
-<?xml version="1.0"?>
-<!DOCTYPE target SYSTEM "gdb-target.dtd">
-<target>
-  <architecture>riscv:rv32</architecture>
-
-  <feature name="org.gnu.gdb.riscv.rv32i">
-    <reg name="x0"  bitsize="32" group="general"/>
-    <reg name="x1"  bitsize="32" group="general"/>
-    <reg name="x2"  bitsize="32" group="general"/>
-    <reg name="x3"  bitsize="32" group="general"/>
-    <reg name="x4"  bitsize="32" group="general"/>
-    <reg name="x5"  bitsize="32" group="general"/>
-    <reg name="x6"  bitsize="32" group="general"/>
-    <reg name="x7"  bitsize="32" group="general"/>
-    <reg name="x8"  bitsize="32" group="general"/>
-    <reg name="x9"  bitsize="32" group="general"/>
-    <reg name="x10" bitsize="32" group="general"/>
-    <reg name="x11" bitsize="32" group="general"/>
-    <reg name="x12" bitsize="32" group="general"/>
-    <reg name="x13" bitsize="32" group="general"/>
-    <reg name="x14" bitsize="32" group="general"/>
-    <reg name="x15" bitsize="32" group="general"/>
-    <reg name="x16" bitsize="32" group="general"/>
-    <reg name="x17" bitsize="32" group="general"/>
-    <reg name="x18" bitsize="32" group="general"/>
-    <reg name="x19" bitsize="32" group="general"/>
-    <reg name="x20" bitsize="32" group="general"/>
-    <reg name="x21" bitsize="32" group="general"/>
-    <reg name="x22" bitsize="32" group="general"/>
-    <reg name="x23" bitsize="32" group="general"/>
-    <reg name="x24" bitsize="32" group="general"/>
-    <reg name="x25" bitsize="32" group="general"/>
-    <reg name="x26" bitsize="32" group="general"/>
-    <reg name="x27" bitsize="32" group="general"/>
-    <reg name="x28" bitsize="32" group="general"/>
-    <reg name="x29" bitsize="32" group="general"/>
-    <reg name="x30" bitsize="32" group="general"/>
-    <reg name="x31" bitsize="32" group="general"/>
-  </feature>
-
-</target>
-
- */
 } // namespace debugger
 } // namespace iss
 
-#endif /* _ISS_DEBUGGER_RISCV_TARGET_ADAPTER_H_ */
+#endif /* _ISS_ARCH_DEBUGGER_RISCV_TARGET_ADAPTER_H_ */

src/main.cpp

@@ -69,7 +69,8 @@ int main(int argc, char* argv[]) {
         ("logfile,l", po::value<std::string>(), "Sets default log file.")
         ("disass,d", po::value<std::string>()->implicit_value(""), "Enables disassembly")
         ("gdb-port,g", po::value<unsigned>()->default_value(0), "enable gdb server and specify port to use")
-        ("instructions,i", po::value<uint64_t>()->default_value(std::numeric_limits<uint64_t>::max()), "max. number of instructions to simulate")
+        ("ilimit,i", po::value<uint64_t>()->default_value(std::numeric_limits<uint64_t>::max()), "max. number of instructions to simulate")
+        ("flimit", po::value<uint64_t>()->default_value(std::numeric_limits<uint64_t>::max()), "max. number of fetches to simulate")
         ("reset,r", po::value<std::string>(), "reset address")
         ("dump-ir", "dump the intermediate representation")
         ("elf,f", po::value<std::vector<std::string>>(), "ELF file(s) to load")
@@ -140,7 +141,10 @@ int main(int argc, char* argv[]) {
             std::tie(cpu, vm) = f.create(isa_opt, clim["gdb-port"].as<unsigned>(), &semihosting_cb);
         }
         if(!cpu) {
-            CPPLOG(ERR) << "Could not create cpu for isa " << isa_opt << " and backend " << clim["backend"].as<std::string>() << std::endl;
+            auto list = f.get_names();
+            std::sort(std::begin(list), std::end(list));
+            CPPLOG(ERR) << "Could not create cpu for isa " << isa_opt << " and backend " << clim["backend"].as<std::string>() << "\n"
+                        << "Available implementations (core|platform|backend):\n  - " << util::join(list, "\n  - ") << std::endl;
             return 127;
         }
         if(!vm) {
@@ -202,21 +206,36 @@ int main(int argc, char* argv[]) {
         if(clim.count("elf"))
             for(std::string input : clim["elf"].as<std::vector<std::string>>()) {
                 auto start_addr = vm->get_arch()->load_file(input);
-                if(start_addr.second) // FIXME: this always evaluates to true as load file always returns <sth, true>
+                if(start_addr.second)
                     start_address = start_addr.first;
+                else {
+                    LOG(ERR) << "Error occured while loading file " << input << std::endl;
+                    return 1;
+                }
             }
         for(std::string input : args) {
             auto start_addr = vm->get_arch()->load_file(input); // treat remaining arguments as elf files
-            if(start_addr.second) // FIXME: this always evaluates to true as load file always returns <sth, true>
+            if(start_addr.second)
                 start_address = start_addr.first;
+            else {
+                LOG(ERR) << "Error occured while loading file " << input << std::endl;
+                return 1;
+            }
         }
         if(clim.count("reset")) {
             auto str = clim["reset"].as<std::string>();
             start_address = str.find("0x") == 0 ? std::stoull(str.substr(2), nullptr, 16) : std::stoull(str, nullptr, 10);
         }
         vm->reset(start_address);
-        auto cycles = clim["instructions"].as<uint64_t>();
-        res = vm->start(cycles, dump);
+        auto limit = clim["ilimit"].as<uint64_t>();
+        auto cond = iss::finish_cond_e::JUMP_TO_SELF;
+        if(clim.count("flimit")) {
+            cond = cond | iss::finish_cond_e::FCOUNT_LIMIT;
+            limit = clim["flimit"].as<uint64_t>();
+        } else {
+            cond = cond | iss::finish_cond_e::ICOUNT_LIMIT;
+        }
+        res = vm->start(limit, dump, cond);
 
         auto instr_if = vm->get_arch()->get_instrumentation_if();
         // this assumes a single input file

src/sysc/core_complex.cpp

@@ -42,7 +42,6 @@
 #include <iss/plugin/loader.h>
 #endif
 #include "sc_core_adapter_if.h"
-#include <iss/arch/tgc_mapper.h>
 #include <scc/report.h>
 #include <util/ities.h>
 #include <iostream>
@@ -125,7 +124,7 @@ using vm_ptr = std::unique_ptr<iss::vm_if>;
 
 class core_wrapper {
 public:
-    core_wrapper(core_complex* owner)
+    core_wrapper(core_complex_if* owner)
     : owner(owner) {}
 
     void reset(uint64_t addr) { vm->reset(addr); }
@@ -181,7 +180,7 @@ public:
                                              "SystemC sub-commands: break <time>, print_time"});
     }
 
-    core_complex* const owner;
+    core_complex_if* const owner;
     vm_ptr vm{nullptr};
     sc_cpu_ptr cpu{nullptr};
     iss::debugger::target_adapter_if* tgt_adapter{nullptr};
@@ -197,9 +196,9 @@ struct core_trace {
     scv_tr_handle tr_handle;
 };
 
-SC_HAS_PROCESS(core_complex); // NOLINT
 #ifndef CWR_SYSTEMC
-core_complex::core_complex(sc_module_name const& name)
+template <unsigned int BUSWIDTH>
+core_complex<BUSWIDTH>::core_complex(sc_module_name const& name)
 : sc_module(name)
 , fetch_lut(tlm_dmi_ext())
 , read_lut(tlm_dmi_ext())
@@ -208,7 +207,7 @@ core_complex::core_complex(sc_module_name const& name)
 }
 #endif
 
-void core_complex::init() {
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::init() {
     trc = new core_trace();
     ibus.register_invalidate_direct_mem_ptr([=](uint64_t start, uint64_t end) -> void {
         auto lut_entry = fetch_lut.getEntry(start);
@@ -227,6 +226,7 @@ void core_complex::init() {
         }
     });
 
+    SC_HAS_PROCESS(core_complex<BUSWIDTH>); // NOLINT
     SC_THREAD(run);
     SC_METHOD(rst_cb);
     sensitive << rst_i;
@@ -252,16 +252,16 @@ void core_complex::init() {
 #endif
 }
 
-core_complex::~core_complex() {
+template <unsigned int BUSWIDTH> core_complex<BUSWIDTH>::~core_complex() {
     delete cpu;
     delete trc;
     for(auto* p : plugin_list)
         delete p;
 }
 
-void core_complex::trace(sc_trace_file* trf) const {}
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::trace(sc_trace_file* trf) const {}
 
-void core_complex::before_end_of_elaboration() {
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::before_end_of_elaboration() {
     SCCDEBUG(SCMOD) << "instantiating iss::arch::tgf with " << GET_PROP_VALUE(backend) << " backend";
     // cpu = scc::make_unique<core_wrapper>(this);
     cpu = new core_wrapper(this);
@@ -302,7 +302,7 @@ void core_complex::before_end_of_elaboration() {
     }
 }
 
-void core_complex::start_of_simulation() {
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::start_of_simulation() {
     // quantum_keeper.reset();
     if(GET_PROP_VALUE(elf_file).size() > 0) {
         istringstream is(GET_PROP_VALUE(elf_file));
@@ -325,7 +325,7 @@ void core_complex::start_of_simulation() {
     }
 }
 
-bool core_complex::disass_output(uint64_t pc, const std::string instr_str) {
+template <unsigned int BUSWIDTH> bool core_complex<BUSWIDTH>::disass_output(uint64_t pc, const std::string instr_str) {
     if(trc->m_db == nullptr)
         return false;
     if(trc->tr_handle.is_active())
@@ -339,7 +339,7 @@ bool core_complex::disass_output(uint64_t pc, const std::string instr_str) {
     return true;
 }
 
-void core_complex::forward() {
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::forward() {
 #ifndef CWR_SYSTEMC
     set_clock_period(clk_i.read());
 #else
@@ -348,24 +348,24 @@ void core_complex::forward() {
 #endif
 }
 
-void core_complex::set_clock_period(sc_core::sc_time period) {
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::set_clock_period(sc_core::sc_time period) {
     curr_clk = period;
     if(period == SC_ZERO_TIME)
         cpu->set_interrupt_execution(true);
 }
 
-void core_complex::rst_cb() {
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::rst_cb() {
     if(rst_i.read())
         cpu->set_interrupt_execution(true);
 }
 
-void core_complex::sw_irq_cb() { cpu->local_irq(3, sw_irq_i.read()); }
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::sw_irq_cb() { cpu->local_irq(3, sw_irq_i.read()); }
 
-void core_complex::timer_irq_cb() { cpu->local_irq(7, timer_irq_i.read()); }
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::timer_irq_cb() { cpu->local_irq(7, timer_irq_i.read()); }
 
-void core_complex::ext_irq_cb() { cpu->local_irq(11, ext_irq_i.read()); }
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::ext_irq_cb() { cpu->local_irq(11, ext_irq_i.read()); }
 
-void core_complex::local_irq_cb() {
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::local_irq_cb() {
     for(auto i = 0U; i < local_irq_i.size(); ++i) {
         if(local_irq_i[i].event()) {
             cpu->local_irq(16 + i, local_irq_i[i].read());
@@ -373,7 +373,7 @@ void core_complex::local_irq_cb() {
     }
 }
 
-void core_complex::run() {
+template <unsigned int BUSWIDTH> void core_complex<BUSWIDTH>::run() {
     wait(SC_ZERO_TIME); // separate from elaboration phase
     do {
         wait(SC_ZERO_TIME);
@@ -391,7 +391,7 @@ void core_complex::run() {
     sc_stop();
 }
 
-bool core_complex::read_mem(uint64_t addr, unsigned length, uint8_t* const data, bool is_fetch) {
+template <unsigned int BUSWIDTH> bool core_complex<BUSWIDTH>::read_mem(uint64_t addr, unsigned length, uint8_t* const data, bool is_fetch) {
     auto& dmi_lut = is_fetch ? fetch_lut : read_lut;
     auto lut_entry = dmi_lut.getEntry(addr);
     if(lut_entry.get_granted_access() != tlm::tlm_dmi::DMI_ACCESS_NONE && addr + length <= lut_entry.get_end_address() + 1) {
@@ -449,7 +449,7 @@ bool core_complex::read_mem(uint64_t addr, unsigned length, uint8_t* const data,
     }
 }
 
-bool core_complex::write_mem(uint64_t addr, unsigned length, const uint8_t* const data) {
+template <unsigned int BUSWIDTH> bool core_complex<BUSWIDTH>::write_mem(uint64_t addr, unsigned length, const uint8_t* const data) {
     auto lut_entry = write_lut.getEntry(addr);
     if(lut_entry.get_granted_access() != tlm::tlm_dmi::DMI_ACCESS_NONE && addr + length <= lut_entry.get_end_address() + 1) {
         auto offset = addr - lut_entry.get_start_address();
@@ -497,7 +497,7 @@ bool core_complex::write_mem(uint64_t addr, unsigned length, const uint8_t* cons
     }
 }
 
-bool core_complex::read_mem_dbg(uint64_t addr, unsigned length, uint8_t* const data) {
+template <unsigned int BUSWIDTH> bool core_complex<BUSWIDTH>::read_mem_dbg(uint64_t addr, unsigned length, uint8_t* const data) {
     tlm::tlm_generic_payload gp;
     gp.set_command(tlm::TLM_READ_COMMAND);
     gp.set_address(addr);
@@ -507,7 +507,7 @@ bool core_complex::read_mem_dbg(uint64_t addr, unsigned length, uint8_t* const d
     return dbus->transport_dbg(gp) == length;
 }
 
-bool core_complex::write_mem_dbg(uint64_t addr, unsigned length, const uint8_t* const data) {
+template <unsigned int BUSWIDTH> bool core_complex<BUSWIDTH>::write_mem_dbg(uint64_t addr, unsigned length, const uint8_t* const data) {
     write_buf.resize(length);
     std::copy(data, data + length, write_buf.begin()); // need to copy as TLM does not guarantee data integrity
     tlm::tlm_generic_payload gp;
@@ -518,5 +518,10 @@ bool core_complex::write_mem_dbg(uint64_t addr, unsigned length, const uint8_t*
     gp.set_streaming_width(length);
     return dbus->transport_dbg(gp) == length;
 }
+
+template class core_complex<scc::LT>;
+template class core_complex<32>;
+template class core_complex<64>;
+
 } /* namespace tgfs */
 } /* namespace sysc */

src/sysc/core_complex.h

@@ -33,6 +33,7 @@
 #ifndef _SYSC_CORE_COMPLEX_H_
 #define _SYSC_CORE_COMPLEX_H_
 
+#include <scc/signal_opt_ports.h>
 #include <scc/tick2time.h>
 #include <scc/traceable.h>
 #include <scc/utilities.h>
@@ -40,10 +41,8 @@
 #include <tlm/scc/scv/tlm_rec_initiator_socket.h>
 #ifdef CWR_SYSTEMC
 #include <scmlinc/scml_property.h>
-#define SOCKET_WIDTH 32
 #else
 #include <cci_configuration>
-#define SOCKET_WIDTH scc::LT
 #endif
 #include <memory>
 #include <tlm>
@@ -68,12 +67,35 @@ public:
 namespace tgfs {
 class core_wrapper;
 struct core_trace;
+struct core_complex_if {
 
-class core_complex : public sc_core::sc_module, public scc::traceable {
+    virtual ~core_complex_if() = default;
+
+    virtual bool read_mem(uint64_t addr, unsigned length, uint8_t* const data, bool is_fetch) = 0;
+
+    virtual bool write_mem(uint64_t addr, unsigned length, const uint8_t* const data) = 0;
+
+    virtual bool read_mem_dbg(uint64_t addr, unsigned length, uint8_t* const data) = 0;
+
+    virtual bool write_mem_dbg(uint64_t addr, unsigned length, const uint8_t* const data) = 0;
+
+    virtual bool disass_output(uint64_t pc, const std::string instr) = 0;
+
+    virtual unsigned get_last_bus_cycles() = 0;
+
+    //! Allow quantum keeper handling
+    virtual void sync(uint64_t) = 0;
+
+    virtual char const* hier_name() = 0;
+
+    scc::sc_in_opt<uint64_t> mtime_i{"mtime_i"};
+};
+
+template <unsigned int BUSWIDTH = scc::LT> class core_complex : public sc_core::sc_module, public scc::traceable, public core_complex_if {
 public:
-    tlm::scc::initiator_mixin<tlm::tlm_initiator_socket<SOCKET_WIDTH>> ibus{"ibus"};
+    tlm::scc::initiator_mixin<tlm::tlm_initiator_socket<BUSWIDTH>> ibus{"ibus"};
 
-    tlm::scc::initiator_mixin<tlm::tlm_initiator_socket<SOCKET_WIDTH>> dbus{"dbus"};
+    tlm::scc::initiator_mixin<tlm::tlm_initiator_socket<BUSWIDTH>> dbus{"dbus"};
 
     sc_core::sc_in<bool> rst_i{"rst_i"};
 
@@ -88,8 +110,6 @@ public:
 #ifndef CWR_SYSTEMC
     sc_core::sc_in<sc_core::sc_time> clk_i{"clk_i"};
 
-    sc_core::sc_port<tlm::tlm_peek_if<uint64_t>, 1, sc_core::SC_ZERO_OR_MORE_BOUND> mtime_o{"mtime_o"};
-
     cci::cci_param<std::string> elf_file{"elf_file", ""};
 
     cci::cci_param<bool> enable_disass{"enable_disass", false};
@@ -115,8 +135,6 @@ public:
 #else
     sc_core::sc_in<bool> clk_i{"clk_i"};
 
-    sc_core::sc_in<uint64_t> mtime_i{"mtime_i"};
-
     scml_property<std::string> elf_file{"elf_file", ""};
 
     scml_property<bool> enable_disass{"enable_disass", false};
@@ -159,13 +177,13 @@ public:
 
     ~core_complex();
 
-    inline unsigned get_last_bus_cycles() {
+    unsigned get_last_bus_cycles() override {
         auto mem_incr = std::max(ibus_inc, dbus_inc);
         ibus_inc = dbus_inc = 0;
         return mem_incr > 1 ? mem_incr : 1;
     }
 
-    inline void sync(uint64_t cycle) {
+    void sync(uint64_t cycle) override {
         auto core_inc = curr_clk * (cycle - last_sync_cycle);
         quantum_keeper.inc(core_inc);
         if(quantum_keeper.need_sync()) {
@@ -175,20 +193,22 @@ public:
         last_sync_cycle = cycle;
     }
 
-    bool read_mem(uint64_t addr, unsigned length, uint8_t* const data, bool is_fetch);
+    bool read_mem(uint64_t addr, unsigned length, uint8_t* const data, bool is_fetch) override;
 
-    bool write_mem(uint64_t addr, unsigned length, const uint8_t* const data);
+    bool write_mem(uint64_t addr, unsigned length, const uint8_t* const data) override;
 
-    bool read_mem_dbg(uint64_t addr, unsigned length, uint8_t* const data);
+    bool read_mem_dbg(uint64_t addr, unsigned length, uint8_t* const data) override;
 
-    bool write_mem_dbg(uint64_t addr, unsigned length, const uint8_t* const data);
+    bool write_mem_dbg(uint64_t addr, unsigned length, const uint8_t* const data) override;
 
     void trace(sc_core::sc_trace_file* trf) const override;
 
-    bool disass_output(uint64_t pc, const std::string instr);
+    bool disass_output(uint64_t pc, const std::string instr) override;
 
     void set_clock_period(sc_core::sc_time period);
 
+    char const* hier_name() override { return name(); }
+
 protected:
     void before_end_of_elaboration() override;
     void start_of_simulation() override;

src/sysc/register_tgc_c.cpp

@@ -46,12 +46,12 @@ using namespace sysc;
 volatile std::array<bool, 2> tgc_init = {
     iss_factory::instance().register_creator("tgc5c|m_p|interp",
                                              [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-                                                 auto cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+                                                 auto cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
                                                  auto* cpu = new sc_core_adapter<arch::riscv_hart_m_p<arch::tgc5c>>(cc);
                                                  return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::tgc5c*>(cpu), gdb_port)}};
                                              }),
     iss_factory::instance().register_creator("tgc5c|mu_p|interp", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-        auto cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+        auto cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
         auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::tgc5c>>(cc);
         return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::tgc5c*>(cpu), gdb_port)}};
     })};
@@ -62,12 +62,12 @@ using namespace sysc;
 volatile std::array<bool, 2> tgc_init = {
     iss_factory::instance().register_creator("tgc5c|m_p|llvm",
                                              [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-                                                 auto cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+                                                 auto cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
                                                  auto* cpu = new sc_core_adapter<arch::riscv_hart_m_p<arch::tgc5c>>(cc);
                                                  return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::tgc5c*>(cpu), gdb_port)}};
                                              }),
     iss_factory::instance().register_creator("tgc5c|mu_p|llvm", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-        auto cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+        auto cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
         auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::tgc5c>>(cc);
         return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::tgc5c*>(cpu), gdb_port)}};
     })};
@@ -79,12 +79,12 @@ using namespace sysc;
 volatile std::array<bool, 2> tgc_init = {
     iss_factory::instance().register_creator("tgc5c|m_p|tcc",
                                              [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-                                                 auto cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+                                                 auto cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
                                                  auto* cpu = new sc_core_adapter<arch::riscv_hart_m_p<arch::tgc5c>>(cc);
                                                  return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::tgc5c*>(cpu), gdb_port)}};
                                              }),
     iss_factory::instance().register_creator("tgc5c|mu_p|tcc", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-        auto cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+        auto cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
         auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::tgc5c>>(cc);
         return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::tgc5c*>(cpu), gdb_port)}};
     })};
@@ -96,12 +96,12 @@ using namespace sysc;
 volatile std::array<bool, 2> tgc_init = {
     iss_factory::instance().register_creator("tgc5c|m_p|asmjit",
                                              [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-                                                 auto cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+                                                 auto cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
                                                  auto* cpu = new sc_core_adapter<arch::riscv_hart_m_p<arch::tgc5c>>(cc);
                                                  return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::tgc5c*>(cpu), gdb_port)}};
                                              }),
     iss_factory::instance().register_creator("tgc5c|mu_p|asmjit", [](unsigned gdb_port, void* data) -> iss_factory::base_t {
-        auto cc = reinterpret_cast<sysc::tgfs::core_complex*>(data);
+        auto cc = reinterpret_cast<sysc::tgfs::core_complex_if*>(data);
         auto* cpu = new sc_core_adapter<arch::riscv_hart_mu_p<arch::tgc5c>>(cc);
         return {sysc::sc_cpu_ptr{cpu}, vm_ptr{create(static_cast<arch::tgc5c*>(cpu), gdb_port)}};
     })};

src/sysc/sc_core_adapter.h

@@ -21,7 +21,7 @@ public:
     using reg_t = typename iss::arch::traits<typename PLAT::core>::reg_t;
     using phys_addr_t = typename iss::arch::traits<typename PLAT::core>::phys_addr_t;
     using heart_state_t = typename PLAT::hart_state_type;
-    sc_core_adapter(sysc::tgfs::core_complex* owner)
+    sc_core_adapter(sysc::tgfs::core_complex_if* owner)
     : owner(owner) {}
 
     iss::arch_if* get_arch_if() override { return this; }
@@ -54,9 +54,9 @@ public:
             std::stringstream s;
             s << "[p:" << lvl[this->reg.PRIV] << ";s:0x" << std::hex << std::setfill('0') << std::setw(sizeof(reg_t) * 2)
               << (reg_t)this->state.mstatus << std::dec << ";c:" << this->reg.icount + this->cycle_offset << "]";
-            SCCDEBUG(owner->name()) << "disass: "
-                                    << "0x" << std::setw(16) << std::right << std::setfill('0') << std::hex << pc << "\t\t" << std::setw(40)
-                                    << std::setfill(' ') << std::left << instr << s.str();
+            SCCDEBUG(owner->hier_name()) << "disass: "
+                                         << "0x" << std::setw(16) << std::right << std::setfill('0') << std::hex << pc << "\t\t"
+                                         << std::setw(40) << std::setfill(' ') << std::left << instr << s.str();
         }
     };
 
@@ -79,10 +79,10 @@ public:
                     switch(hostvar >> 48) {
                     case 0:
                         if(hostvar != 0x1) {
-                            SCCINFO(owner->name())
+                            SCCINFO(owner->hier_name())
                                 << "tohost value is 0x" << std::hex << hostvar << std::dec << " (" << hostvar << "), stopping simulation";
                         } else {
-                            SCCINFO(owner->name())
+                            SCCINFO(owner->hier_name())
                                 << "tohost value is 0x" << std::hex << hostvar << std::dec << " (" << hostvar << "), stopping simulation";
                         }
                         this->reg.trap_state = std::numeric_limits<uint32_t>::max();
@@ -112,21 +112,8 @@ public:
     }
 
     iss::status read_csr(unsigned addr, reg_t& val) override {
-#ifndef CWR_SYSTEMC
-        if((addr == iss::arch::time || addr == iss::arch::timeh) && owner->mtime_o.get_interface(0)) {
-            uint64_t time_val;
-            bool ret = owner->mtime_o->nb_peek(time_val);
-            if(addr == iss::arch::time) {
-                val = static_cast<reg_t>(time_val);
-            } else if(addr == iss::arch::timeh) {
-                if(sizeof(reg_t) != 4)
-                    return iss::Err;
-                val = static_cast<reg_t>(time_val >> 32);
-            }
-            return ret ? iss::Ok : iss::Err;
-#else
         if((addr == iss::arch::time || addr == iss::arch::timeh)) {
-            uint64_t time_val = owner->mtime_i.read();
+            uint64_t time_val = owner->mtime_i.get_interface() ? owner->mtime_i.read() : 0;
             if(addr == iss::arch::time) {
                 val = static_cast<reg_t>(time_val);
             } else if(addr == iss::arch::timeh) {
@@ -135,14 +122,13 @@ public:
                 val = static_cast<reg_t>(time_val >> 32);
             }
             return iss::Ok;
-#endif
         } else {
             return PLAT::read_csr(addr, val);
         }
     }
 
     void wait_until(uint64_t flags) override {
-        SCCDEBUG(owner->name()) << "Sleeping until interrupt";
+        SCCDEBUG(owner->hier_name()) << "Sleeping until interrupt";
         while(this->reg.pending_trap == 0 && (this->csr[iss::arch::mip] & this->csr[iss::arch::mie]) == 0) {
             sc_core::wait(wfi_evt);
         }
@@ -173,11 +159,11 @@ public:
             this->csr[iss::arch::mip] &= ~mask;
         this->check_interrupt();
         if(value)
-            SCCTRACE(owner->name()) << "Triggering interrupt " << id << " Pending trap: " << this->reg.pending_trap;
+            SCCTRACE(owner->hier_name()) << "Triggering interrupt " << id << " Pending trap: " << this->reg.pending_trap;
     }
 
 private:
-    sysc::tgfs::core_complex* const owner;
+    sysc::tgfs::core_complex_if* const owner{nullptr};
     sc_core::sc_event wfi_evt;
     uint64_t hostvar{std::numeric_limits<uint64_t>::max()};
     unsigned to_host_wr_cnt = 0;

src/vm/asmjit/vm_tgc5c.cpp

@@ -88,14 +88,13 @@ protected:
     using super::write_reg_to_mem;
     using super::gen_read_mem;
     using super::gen_write_mem;
-    using super::gen_wait;
     using super::gen_leave;
     using super::gen_sync;
    
     using this_class = vm_impl<ARCH>;
     using compile_func = continuation_e (this_class::*)(virt_addr_t&, code_word_t, jit_holder&);
 
-    continuation_e gen_single_inst_behavior(virt_addr_t&, unsigned int &, jit_holder&) override;
+    continuation_e gen_single_inst_behavior(virt_addr_t&, jit_holder&) override;
     enum globals_e {TVAL = 0, GLOBALS_SIZE};
     void gen_block_prologue(jit_holder& jh) override;
     void gen_block_epilogue(jit_holder& jh) override;
@@ -104,7 +103,7 @@ protected:
     void gen_instr_prologue(jit_holder& jh);
     void gen_instr_epilogue(jit_holder& jh);
     inline void gen_raise(jit_holder& jh, uint16_t trap_id, uint16_t cause);
-    template <typename T, typename = std::enable_if_t<std::is_integral_v<T>>> void gen_set_tval(jit_holder& jh, T new_tval) ;
+    template <typename T, typename = typename std::enable_if<std::is_integral<T>::value>::type> void gen_set_tval(jit_holder& jh, T new_tval) ;
     void gen_set_tval(jit_holder& jh, x86_reg_t _new_tval) ;
 
     template<unsigned W, typename U, typename S = typename std::make_signed<U>::type>
@@ -112,7 +111,8 @@ protected:
         auto mask = (1ULL<<W) - 1;
         auto sign_mask = 1ULL<<(W-1);
         return (from & mask) | ((from & sign_mask) ? ~mask : 0);
-    } 
+    }
+
 private:
     /****************************************************************************
      * start opcode definitions
@@ -500,6 +500,7 @@ private:
                 (gen_operation(cc, band, (gen_operation(cc, add, load_reg_from_mem(jh, traits::X0 + rs1), (int16_t)sext<12>(imm))
                 ), addr_mask)
                 ), 32, true);
+            {
             auto label_merge = cc.newLabel();
             cmp(cc, gen_operation(cc, urem, new_pc, static_cast<uint32_t>(traits::INSTR_ALIGNMENT))
             ,0);
@@ -521,6 +522,7 @@ private:
                     mov(cc, get_ptr_for(jh, traits::LAST_BRANCH), static_cast<int>(UNKNOWN_JUMP));
                 }
             cc.bind(label_merge);
+            }
         }
         auto returnValue = BRANCH;
         
@@ -566,6 +568,7 @@ private:
             gen_raise(jh, 0, static_cast<int32_t>(traits::RV_CAUSE_ILLEGAL_INSTRUCTION));
         }
         else{
+            {
             auto label_merge = cc.newLabel();
             cmp(cc, gen_operation(cc, eq, load_reg_from_mem(jh, traits::X0 + rs1), load_reg_from_mem(jh, traits::X0 + rs2))
             ,0);
@@ -583,6 +586,7 @@ private:
                 }
             }
             cc.bind(label_merge);
+            }
         }
         auto returnValue = BRANCH;
         
@@ -628,6 +632,7 @@ private:
             gen_raise(jh, 0, static_cast<int32_t>(traits::RV_CAUSE_ILLEGAL_INSTRUCTION));
         }
         else{
+            {
             auto label_merge = cc.newLabel();
             cmp(cc, gen_operation(cc, ne, load_reg_from_mem(jh, traits::X0 + rs1), load_reg_from_mem(jh, traits::X0 + rs2))
             ,0);
@@ -645,6 +650,7 @@ private:
                 }
             }
             cc.bind(label_merge);
+            }
         }
         auto returnValue = BRANCH;
         
@@ -690,6 +696,7 @@ private:
             gen_raise(jh, 0, static_cast<int32_t>(traits::RV_CAUSE_ILLEGAL_INSTRUCTION));
         }
         else{
+            {
             auto label_merge = cc.newLabel();
             cmp(cc, gen_operation(cc, lt, gen_ext(cc, 
                 load_reg_from_mem(jh, traits::X0 + rs1), 32, false), gen_ext(cc, 
@@ -709,6 +716,7 @@ private:
                 }
             }
             cc.bind(label_merge);
+            }
         }
         auto returnValue = BRANCH;
         
@@ -754,6 +762,7 @@ private:
             gen_raise(jh, 0, static_cast<int32_t>(traits::RV_CAUSE_ILLEGAL_INSTRUCTION));
         }
         else{
+            {
             auto label_merge = cc.newLabel();
             cmp(cc, gen_operation(cc, gte, gen_ext(cc, 
                 load_reg_from_mem(jh, traits::X0 + rs1), 32, false), gen_ext(cc, 
@@ -773,6 +782,7 @@ private:
                 }
             }
             cc.bind(label_merge);
+            }
         }
         auto returnValue = BRANCH;
         
@@ -818,6 +828,7 @@ private:
             gen_raise(jh, 0, static_cast<int32_t>(traits::RV_CAUSE_ILLEGAL_INSTRUCTION));
         }
         else{
+            {
             auto label_merge = cc.newLabel();
             cmp(cc, gen_operation(cc, ltu, load_reg_from_mem(jh, traits::X0 + rs1), load_reg_from_mem(jh, traits::X0 + rs2))
             ,0);
@@ -835,6 +846,7 @@ private:
                 }
             }
             cc.bind(label_merge);
+            }
         }
         auto returnValue = BRANCH;
         
@@ -880,6 +892,7 @@ private:
             gen_raise(jh, 0, static_cast<int32_t>(traits::RV_CAUSE_ILLEGAL_INSTRUCTION));
         }
         else{
+            {
             auto label_merge = cc.newLabel();
             cmp(cc, gen_operation(cc, gteu, load_reg_from_mem(jh, traits::X0 + rs1), load_reg_from_mem(jh, traits::X0 + rs2))
             ,0);
@@ -897,6 +910,7 @@ private:
                 }
             }
             cc.bind(label_merge);
+            }
         }
         auto returnValue = BRANCH;
         
@@ -1407,23 +1421,21 @@ private:
         }
         else{
             if(rd!=0){
-                {
-                auto label_then = cc.newLabel();
-                auto label_merge = cc.newLabel();
-                auto tmp_reg = get_reg_for(cc, 1);
+                auto label_then11 = cc.newLabel();
+                auto label_merge11 = cc.newLabel();
+                auto tmp_reg11 = get_reg(cc, 8, false);
                 cmp(cc, gen_ext(cc, 
                     load_reg_from_mem(jh, traits::X0 + rs1), 32, true), (int16_t)sext<12>(imm));
-                cc.jl(label_then);
-                mov(cc, tmp_reg,0);
-                cc.jmp(label_merge);
-                cc.bind(label_then);
-                mov(cc, tmp_reg,1);
-                cc.bind(label_merge);
+                cc.jl(label_then11);
+                mov(cc, tmp_reg11,0);
+                cc.jmp(label_merge11);
+                cc.bind(label_then11);
+                mov(cc, tmp_reg11, 1);
+                cc.bind(label_merge11);
                 mov(cc, get_ptr_for(jh, traits::X0+ rd),
-                      gen_ext(cc, tmp_reg
+                      gen_ext(cc, tmp_reg11
                       , 32, false)
                 );
-                }
             }
         }
         auto returnValue = CONT;
@@ -1470,22 +1482,20 @@ private:
         }
         else{
             if(rd!=0){
-                {
-                auto label_then = cc.newLabel();
-                auto label_merge = cc.newLabel();
-                auto tmp_reg = get_reg_for(cc, 1);
+                auto label_then12 = cc.newLabel();
+                auto label_merge12 = cc.newLabel();
+                auto tmp_reg12 = get_reg(cc, 8, false);
                 cmp(cc, load_reg_from_mem(jh, traits::X0 + rs1), (uint32_t)((int16_t)sext<12>(imm)));
-                cc.jb(label_then);
-                mov(cc, tmp_reg,0);
-                cc.jmp(label_merge);
-                cc.bind(label_then);
-                mov(cc, tmp_reg,1);
-                cc.bind(label_merge);
+                cc.jb(label_then12);
+                mov(cc, tmp_reg12,0);
+                cc.jmp(label_merge12);
+                cc.bind(label_then12);
+                mov(cc, tmp_reg12, 1);
+                cc.bind(label_merge12);
                 mov(cc, get_ptr_for(jh, traits::X0+ rd),
-                      gen_ext(cc, tmp_reg
+                      gen_ext(cc, tmp_reg12
                       , 32, false)
                 );
-                }
             }
         }
         auto returnValue = CONT;
@@ -1978,24 +1988,22 @@ private:
         }
         else{
             if(rd!=0){
-                {
-                auto label_then = cc.newLabel();
-                auto label_merge = cc.newLabel();
-                auto tmp_reg = get_reg_for(cc, 1);
+                auto label_then13 = cc.newLabel();
+                auto label_merge13 = cc.newLabel();
+                auto tmp_reg13 = get_reg(cc, 8, false);
                 cmp(cc, gen_ext(cc, 
                     load_reg_from_mem(jh, traits::X0 + rs1), 32, true), gen_ext(cc, 
                     load_reg_from_mem(jh, traits::X0 + rs2), 32, true));
-                cc.jl(label_then);
-                mov(cc, tmp_reg,0);
-                cc.jmp(label_merge);
-                cc.bind(label_then);
-                mov(cc, tmp_reg,1);
-                cc.bind(label_merge);
+                cc.jl(label_then13);
+                mov(cc, tmp_reg13,0);
+                cc.jmp(label_merge13);
+                cc.bind(label_then13);
+                mov(cc, tmp_reg13, 1);
+                cc.bind(label_merge13);
                 mov(cc, get_ptr_for(jh, traits::X0+ rd),
-                      gen_ext(cc, tmp_reg
+                      gen_ext(cc, tmp_reg13
                       , 32, false)
                 );
-                }
             }
         }
         auto returnValue = CONT;
@@ -2042,22 +2050,20 @@ private:
         }
         else{
             if(rd!=0){
-                {
-                auto label_then = cc.newLabel();
-                auto label_merge = cc.newLabel();
-                auto tmp_reg = get_reg_for(cc, 1);
+                auto label_then14 = cc.newLabel();
+                auto label_merge14 = cc.newLabel();
+                auto tmp_reg14 = get_reg(cc, 8, false);
                 cmp(cc, load_reg_from_mem(jh, traits::X0 + rs1), load_reg_from_mem(jh, traits::X0 + rs2));
-                cc.jb(label_then);
-                mov(cc, tmp_reg,0);
-                cc.jmp(label_merge);
-                cc.bind(label_then);
-                mov(cc, tmp_reg,1);
-                cc.bind(label_merge);
+                cc.jb(label_then14);
+                mov(cc, tmp_reg14,0);
+                cc.jmp(label_merge14);
+                cc.bind(label_then14);
+                mov(cc, tmp_reg14, 1);
+                cc.bind(label_merge14);
                 mov(cc, get_ptr_for(jh, traits::X0+ rd),
-                      gen_ext(cc, tmp_reg
+                      gen_ext(cc, tmp_reg14
                       , 32, false)
                 );
-                }
             }
         }
         auto returnValue = CONT;
@@ -2364,7 +2370,7 @@ private:
         if(this->disass_enabled){
             /* generate disass */
             
-            //This disass is not yet implemented
+            //No disass specified, using instruction name
             std::string mnemonic = "ecall";
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -2401,7 +2407,7 @@ private:
         if(this->disass_enabled){
             /* generate disass */
             
-            //This disass is not yet implemented
+            //No disass specified, using instruction name
             std::string mnemonic = "ebreak";
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -2438,7 +2444,7 @@ private:
         if(this->disass_enabled){
             /* generate disass */
             
-            //This disass is not yet implemented
+            //No disass specified, using instruction name
             std::string mnemonic = "mret";
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -2475,7 +2481,7 @@ private:
         if(this->disass_enabled){
             /* generate disass */
             
-            //This disass is not yet implemented
+            //No disass specified, using instruction name
             std::string mnemonic = "wfi";
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -2497,7 +2503,10 @@ private:
         gen_instr_prologue(jh);
         cc.comment("//behavior:");
         /*generate behavior*/
-        gen_wait(jh, 1);
+        InvokeNode* call_wait_15;
+        jh.cc.comment("//call_wait");
+        jh.cc.invoke(&call_wait_15, &wait, FuncSignature::build<void, int32_t>());
+        setArg(call_wait_15, 0, 1);
         auto returnValue = CONT;
         
         gen_sync(jh, POST_SYNC, 41);
@@ -3116,6 +3125,7 @@ private:
             auto divisor = gen_ext(cc, 
                 load_reg_from_mem(jh, traits::X0 + rs2), 32, true);
             if(rd!=0){
+                {
                 auto label_merge = cc.newLabel();
                 cmp(cc, gen_operation(cc, ne, divisor, 0)
                 ,0);
@@ -3123,6 +3133,7 @@ private:
                 cc.je(label_else);
                 {
                     auto MMIN = ((uint32_t)1)<<(static_cast<uint32_t>(traits::XLEN)-1);
+                    {
                     auto label_merge = cc.newLabel();
                     cmp(cc, gen_operation(cc, land, gen_operation(cc, eq, load_reg_from_mem(jh, traits::X0 + rs1), MMIN)
                     , gen_operation(cc, eq, divisor, - 1)
@@ -3143,6 +3154,7 @@ private:
                                       ), 32, true));
                         }
                     cc.bind(label_merge);
+                    }
                 }
                 cc.jmp(label_merge);
                 cc.bind(label_else);
@@ -3151,6 +3163,7 @@ private:
                               (uint32_t)- 1);
                     }
                 cc.bind(label_merge);
+                }
             }
         }
         auto returnValue = CONT;
@@ -3196,6 +3209,7 @@ private:
             gen_raise(jh, 0, static_cast<int32_t>(traits::RV_CAUSE_ILLEGAL_INSTRUCTION));
         }
         else{
+            {
             auto label_merge = cc.newLabel();
             cmp(cc, gen_operation(cc, ne, load_reg_from_mem(jh, traits::X0 + rs2), 0)
             ,0);
@@ -3217,6 +3231,7 @@ private:
                     }
                 }
             cc.bind(label_merge);
+            }
         }
         auto returnValue = CONT;
         
@@ -3261,6 +3276,7 @@ private:
             gen_raise(jh, 0, static_cast<int32_t>(traits::RV_CAUSE_ILLEGAL_INSTRUCTION));
         }
         else{
+            {
             auto label_merge = cc.newLabel();
             cmp(cc, gen_operation(cc, ne, load_reg_from_mem(jh, traits::X0 + rs2), 0)
             ,0);
@@ -3268,6 +3284,7 @@ private:
             cc.je(label_else);
             {
                 auto MMIN = (uint32_t)1<<(static_cast<uint32_t>(traits::XLEN)-1);
+                {
                 auto label_merge = cc.newLabel();
                 cmp(cc, gen_operation(cc, land, gen_operation(cc, eq, load_reg_from_mem(jh, traits::X0 + rs1), MMIN)
                 , gen_operation(cc, eq, gen_ext(cc, 
@@ -3296,6 +3313,7 @@ private:
                         }
                     }
                 cc.bind(label_merge);
+                }
             }
             cc.jmp(label_merge);
             cc.bind(label_else);
@@ -3306,6 +3324,7 @@ private:
                     }
                 }
             cc.bind(label_merge);
+            }
         }
         auto returnValue = CONT;
         
@@ -3350,6 +3369,7 @@ private:
             gen_raise(jh, 0, static_cast<int32_t>(traits::RV_CAUSE_ILLEGAL_INSTRUCTION));
         }
         else{
+            {
             auto label_merge = cc.newLabel();
             cmp(cc, gen_operation(cc, ne, load_reg_from_mem(jh, traits::X0 + rs2), 0)
             ,0);
@@ -3371,6 +3391,7 @@ private:
                     }
                 }
             cc.bind(label_merge);
+            }
         }
         auto returnValue = CONT;
         
@@ -3388,7 +3409,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, {imm:#05x}", fmt::arg("mnemonic", "c__addi4spn"),
+                "{mnemonic:10} {rd}, {imm:#05x}", fmt::arg("mnemonic", "c.addi4spn"),
                 fmt::arg("rd", name(8+rd)), fmt::arg("imm", imm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3436,7 +3457,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, {uimm:#05x}({rs1})", fmt::arg("mnemonic", "c__lw"),
+                "{mnemonic:10} {rd}, {uimm:#05x}({rs1})", fmt::arg("mnemonic", "c.lw"),
                 fmt::arg("rd", name(8+rd)), fmt::arg("uimm", uimm), fmt::arg("rs1", name(8+rs1)));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3482,7 +3503,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs2}, {uimm:#05x}({rs1})", fmt::arg("mnemonic", "c__sw"),
+                "{mnemonic:10} {rs2}, {uimm:#05x}({rs1})", fmt::arg("mnemonic", "c.sw"),
                 fmt::arg("rs2", name(8+rs2)), fmt::arg("uimm", uimm), fmt::arg("rs1", name(8+rs1)));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3525,7 +3546,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs1}, {imm:#05x}", fmt::arg("mnemonic", "c__addi"),
+                "{mnemonic:10} {rs1}, {imm:#05x}", fmt::arg("mnemonic", "c.addi"),
                 fmt::arg("rs1", name(rs1)), fmt::arg("imm", imm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3572,8 +3593,8 @@ private:
         if(this->disass_enabled){
             /* generate disass */
             
-            //This disass is not yet implemented
-            std::string mnemonic = "c__nop";
+            //No disass specified, using instruction name
+            std::string mnemonic = "c.nop";
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
             jh.disass_collection.push_back(mnemonic_ptr);
@@ -3609,7 +3630,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {imm:#05x}", fmt::arg("mnemonic", "c__jal"),
+                "{mnemonic:10} {imm:#05x}", fmt::arg("mnemonic", "c.jal"),
                 fmt::arg("imm", imm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3653,7 +3674,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, {imm:#05x}", fmt::arg("mnemonic", "c__li"),
+                "{mnemonic:10} {rd}, {imm:#05x}", fmt::arg("mnemonic", "c.li"),
                 fmt::arg("rd", name(rd)), fmt::arg("imm", imm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3700,7 +3721,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, {imm:#05x}", fmt::arg("mnemonic", "c__lui"),
+                "{mnemonic:10} {rd}, {imm:#05x}", fmt::arg("mnemonic", "c.lui"),
                 fmt::arg("rd", name(rd)), fmt::arg("imm", imm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3744,7 +3765,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {nzimm:#05x}", fmt::arg("mnemonic", "c__addi16sp"),
+                "{mnemonic:10} {nzimm:#05x}", fmt::arg("mnemonic", "c.addi16sp"),
                 fmt::arg("nzimm", nzimm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3789,8 +3810,8 @@ private:
         if(this->disass_enabled){
             /* generate disass */
             
-            //This disass is not yet implemented
-            std::string mnemonic = "__reserved_clui";
+            //No disass specified, using instruction name
+            std::string mnemonic = ".reserved_clui";
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
             jh.disass_collection.push_back(mnemonic_ptr);
@@ -3828,7 +3849,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs1}, {shamt}", fmt::arg("mnemonic", "c__srli"),
+                "{mnemonic:10} {rs1}, {shamt}", fmt::arg("mnemonic", "c.srli"),
                 fmt::arg("rs1", name(8+rs1)), fmt::arg("shamt", shamt));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3869,7 +3890,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs1}, {shamt}", fmt::arg("mnemonic", "c__srai"),
+                "{mnemonic:10} {rs1}, {shamt}", fmt::arg("mnemonic", "c.srai"),
                 fmt::arg("rs1", name(8+rs1)), fmt::arg("shamt", shamt));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3923,7 +3944,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs1}, {imm:#05x}", fmt::arg("mnemonic", "c__andi"),
+                "{mnemonic:10} {rs1}, {imm:#05x}", fmt::arg("mnemonic", "c.andi"),
                 fmt::arg("rs1", name(8+rs1)), fmt::arg("imm", imm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -3965,7 +3986,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c__sub"),
+                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c.sub"),
                 fmt::arg("rd", name(8+rd)), fmt::arg("rs2", name(8+rs2)));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4007,7 +4028,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c__xor"),
+                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c.xor"),
                 fmt::arg("rd", name(8+rd)), fmt::arg("rs2", name(8+rs2)));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4048,7 +4069,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c__or"),
+                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c.or"),
                 fmt::arg("rd", name(8+rd)), fmt::arg("rs2", name(8+rs2)));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4089,7 +4110,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c__and"),
+                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c.and"),
                 fmt::arg("rd", name(8+rd)), fmt::arg("rs2", name(8+rs2)));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4129,7 +4150,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {imm:#05x}", fmt::arg("mnemonic", "c__j"),
+                "{mnemonic:10} {imm:#05x}", fmt::arg("mnemonic", "c.j"),
                 fmt::arg("imm", imm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4171,7 +4192,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs1}, {imm:#05x}", fmt::arg("mnemonic", "c__beqz"),
+                "{mnemonic:10} {rs1}, {imm:#05x}", fmt::arg("mnemonic", "c.beqz"),
                 fmt::arg("rs1", name(8+rs1)), fmt::arg("imm", imm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4194,6 +4215,7 @@ private:
         cc.comment("//behavior:");
         /*generate behavior*/
         mov(jh.cc, get_ptr_for(jh, traits::LAST_BRANCH), static_cast<int>(NO_JUMP));
+        {
         auto label_merge = cc.newLabel();
         cmp(cc, gen_operation(cc, eq, load_reg_from_mem(jh, traits::X0 + rs1+8), 0)
         ,0);
@@ -4204,6 +4226,7 @@ private:
             mov(cc, get_ptr_for(jh, traits::LAST_BRANCH), static_cast<int>(KNOWN_JUMP));
         }
         cc.bind(label_merge);
+        }
         auto returnValue = BRANCH;
         
         gen_sync(jh, POST_SYNC, 75);
@@ -4220,7 +4243,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs1}, {imm:#05x}", fmt::arg("mnemonic", "c__bnez"),
+                "{mnemonic:10} {rs1}, {imm:#05x}", fmt::arg("mnemonic", "c.bnez"),
                 fmt::arg("rs1", name(8+rs1)), fmt::arg("imm", imm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4243,6 +4266,7 @@ private:
         cc.comment("//behavior:");
         /*generate behavior*/
         mov(jh.cc, get_ptr_for(jh, traits::LAST_BRANCH), static_cast<int>(NO_JUMP));
+        {
         auto label_merge = cc.newLabel();
         cmp(cc, gen_operation(cc, ne, load_reg_from_mem(jh, traits::X0 + rs1+8), 0)
         ,0);
@@ -4253,6 +4277,7 @@ private:
             mov(cc, get_ptr_for(jh, traits::LAST_BRANCH), static_cast<int>(KNOWN_JUMP));
         }
         cc.bind(label_merge);
+        }
         auto returnValue = BRANCH;
         
         gen_sync(jh, POST_SYNC, 76);
@@ -4269,7 +4294,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs1}, {nzuimm}", fmt::arg("mnemonic", "c__slli"),
+                "{mnemonic:10} {rs1}, {nzuimm}", fmt::arg("mnemonic", "c.slli"),
                 fmt::arg("rs1", name(rs1)), fmt::arg("nzuimm", nzuimm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4317,7 +4342,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, sp, {uimm:#05x}", fmt::arg("mnemonic", "c__lwsp"),
+                "{mnemonic:10} {rd}, sp, {uimm:#05x}", fmt::arg("mnemonic", "c.lwsp"),
                 fmt::arg("rd", name(rd)), fmt::arg("uimm", uimm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4367,7 +4392,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c__mv"),
+                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c.mv"),
                 fmt::arg("rd", name(rd)), fmt::arg("rs2", name(rs2)));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4413,7 +4438,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs1}", fmt::arg("mnemonic", "c__jr"),
+                "{mnemonic:10} {rs1}", fmt::arg("mnemonic", "c.jr"),
                 fmt::arg("rs1", name(rs1)));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4459,8 +4484,8 @@ private:
         if(this->disass_enabled){
             /* generate disass */
             
-            //This disass is not yet implemented
-            std::string mnemonic = "__reserved_cmv";
+            //No disass specified, using instruction name
+            std::string mnemonic = ".reserved_cmv";
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
             jh.disass_collection.push_back(mnemonic_ptr);
@@ -4498,7 +4523,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c__add"),
+                "{mnemonic:10} {rd}, {rs2}", fmt::arg("mnemonic", "c.add"),
                 fmt::arg("rd", name(rd)), fmt::arg("rs2", name(rs2)));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4546,7 +4571,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs1}", fmt::arg("mnemonic", "c__jalr"),
+                "{mnemonic:10} {rs1}", fmt::arg("mnemonic", "c.jalr"),
                 fmt::arg("rs1", name(rs1)));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4595,8 +4620,8 @@ private:
         if(this->disass_enabled){
             /* generate disass */
             
-            //This disass is not yet implemented
-            std::string mnemonic = "c__ebreak";
+            //No disass specified, using instruction name
+            std::string mnemonic = "c.ebreak";
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
             jh.disass_collection.push_back(mnemonic_ptr);
@@ -4634,7 +4659,7 @@ private:
             /* generate disass */
             
             auto mnemonic = fmt::format(
-                "{mnemonic:10} {rs2}, {uimm:#05x}(sp)", fmt::arg("mnemonic", "c__swsp"),
+                "{mnemonic:10} {rs2}, {uimm:#05x}(sp)", fmt::arg("mnemonic", "c.swsp"),
                 fmt::arg("rs2", name(rs2)), fmt::arg("uimm", uimm));
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4679,7 +4704,7 @@ private:
         if(this->disass_enabled){
             /* generate disass */
             
-            //This disass is not yet implemented
+            //No disass specified, using instruction name
             std::string mnemonic = "dii";
             InvokeNode* call_print_disass;
             char* mnemonic_ptr = strdup(mnemonic.c_str());
@@ -4735,7 +4760,7 @@ private:
         gen_raise(jh, 0, 2);
         gen_sync(jh, POST_SYNC, instr_descr.size());
         gen_instr_epilogue(jh);
-        return BRANCH;
+        return ILLEGAL_INSTR;
     }
 };
 
@@ -4755,7 +4780,7 @@ vm_impl<ARCH>::vm_impl(ARCH &core, unsigned core_id, unsigned cluster_id)
     }()) {}
 
 template <typename ARCH>
-continuation_e vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, unsigned int &inst_cnt, jit_holder& jh) {
+continuation_e vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, jit_holder& jh) {
     enum {TRAP_ID=1<<16};
     code_word_t instr = 0;
     phys_addr_t paddr(pc);
@@ -4764,10 +4789,9 @@ continuation_e vm_impl<ARCH>::gen_single_inst_behavior(virt_addr_t &pc, unsigned
         paddr = this->core.virt2phys(pc);
     auto res = this->core.read(paddr, 4, data);
     if (res != iss::Ok)
-        throw trap_access(TRAP_ID, pc.val);
+        return ILLEGAL_FETCH;
     if (instr == 0x0000006f || (instr&0xffff)==0xa001)
-        throw simulation_stopped(0); // 'J 0' or 'C.J 0'
-    ++inst_cnt;
+        return JUMP_TO_SELF;
     uint32_t inst_index = instr_decoder.decode_instr(instr);
     compile_func f = nullptr;
     if(inst_index < instr_descr.size())
@@ -4797,6 +4821,7 @@ void vm_impl<ARCH>::gen_instr_epilogue(jit_holder& jh) {
     cmp(cc, current_trap_state, 0);
     cc.jne(jh.trap_entry);
     cc.inc(get_ptr_for(jh, traits::ICOUNT));
+    cc.inc(get_ptr_for(jh, traits::CYCLE));
 }
 template <typename ARCH>
 void vm_impl<ARCH>::gen_block_prologue(jit_holder& jh){
@@ -4842,6 +4867,7 @@ inline void vm_impl<ARCH>::gen_raise(jit_holder& jh, uint16_t trap_id, uint16_t
     auto tmp1 = get_reg_for(cc, traits::TRAP_STATE);
     mov(cc, tmp1, 0x80ULL << 24 | (cause << 16) | trap_id);
     mov(cc, get_ptr_for(jh, traits::TRAP_STATE), tmp1);
+    cc.jmp(jh.trap_entry);
 }
 template <typename ARCH>
 template <typename T, typename>
@@ -4850,8 +4876,8 @@ void vm_impl<ARCH>::gen_set_tval(jit_holder& jh, T new_tval) {
     }
 template <typename ARCH>
 void vm_impl<ARCH>::gen_set_tval(jit_holder& jh, x86_reg_t _new_tval) {
-    if(std::holds_alternative<x86::Gp>(_new_tval)) {
-        x86::Gp new_tval = std::get<x86::Gp>(_new_tval);
+    if(nonstd::holds_alternative<x86::Gp>(_new_tval)) {
+        x86::Gp new_tval = nonstd::get<x86::Gp>(_new_tval);
         if(new_tval.size() < 8)
             new_tval = gen_ext_Gp(jh.cc, new_tval, 64, false);
         mov(jh.cc, jh.globals[TVAL], new_tval);

src/vm/fp_functions.cpp

@@ -128,7 +128,6 @@ uint32_t fcmp_s(uint32_t v1, uint32_t v2, uint32_t op) {
 }
 
 uint32_t fcvt_s(uint32_t v1, uint32_t op, uint8_t mode) {
-
     float32_t v1f{v1};
     softfloat_exceptionFlags = 0;
     float32_t r;
@@ -204,8 +203,8 @@ uint32_t fclass_s(uint32_t v1) {
     uA.f = a;
     uiA = uA.ui;
 
-    uint_fast16_t infOrNaN = expF32UI(uiA) == 0xFF;
-    uint_fast16_t subnormalOrZero = expF32UI(uiA) == 0;
+    bool infOrNaN = expF32UI(uiA) == 0xFF;
+    bool subnormalOrZero = expF32UI(uiA) == 0;
     bool sign = signF32UI(uiA);
     bool fracZero = fracF32UI(uiA) == 0;
     bool isNaN = isNaNF32UI(uiA);
@@ -218,9 +217,13 @@ uint32_t fclass_s(uint32_t v1) {
 }
 
 uint32_t fconv_d2f(uint64_t v1, uint8_t mode) {
+    bool isNan = isNaNF64UI(v1);
+    bool isSNaN = softfloat_isSigNaNF64UI(v1);
     softfloat_roundingMode = rmm_map.at(mode);
-    bool nan = (v1 & defaultNaNF64UI) == defaultNaNF64UI;
-    if(nan) {
+    softfloat_exceptionFlags = 0;
+    if(isNan) {
+        if(isSNaN)
+            softfloat_raiseFlags(softfloat_flag_invalid);
         return defaultNaNF32UI;
     } else {
         float32_t res = f64_to_f32(float64_t{v1});
@@ -229,11 +232,11 @@ uint32_t fconv_d2f(uint64_t v1, uint8_t mode) {
 }
 
 uint64_t fconv_f2d(uint32_t v1, uint8_t mode) {
-    bool nan = (v1 & defaultNaNF32UI) == defaultNaNF32UI;
-    if(nan) {
+    bool infOrNaN = expF32UI(v1) == 0xFF;
+    bool subnormalOrZero = expF32UI(v1) == 0;
+    if(infOrNaN || subnormalOrZero) {
         return defaultNaNF64UI;
     } else {
-        softfloat_roundingMode = rmm_map.at(mode);
         float64_t res = f32_to_f64(float32_t{v1});
         return res.v;
     }
@@ -313,22 +316,23 @@ uint64_t fcmp_d(uint64_t v1, uint64_t v2, uint32_t op) {
 }
 
 uint64_t fcvt_d(uint64_t v1, uint32_t op, uint8_t mode) {
+
     float64_t v1f{v1};
     softfloat_exceptionFlags = 0;
     float64_t r;
     switch(op) {
-    case 0: { // l->d, fp to int32
+    case 0: { // l from d
         int64_t res = f64_to_i64(v1f, rmm_map.at(mode), true);
         return (uint64_t)res;
     }
-    case 1: { // lu->s
+    case 1: { // lu from d
         uint64_t res = f64_to_ui64(v1f, rmm_map.at(mode), true);
         return res;
     }
-    case 2: // s->l
+    case 2: // d from l
         r = i64_to_f64(v1);
         return r.v;
-    case 3: // s->lu
+    case 3: // d from lu
         r = ui64_to_f64(v1);
         return r.v;
     }
@@ -336,12 +340,24 @@ uint64_t fcvt_d(uint64_t v1, uint32_t op, uint8_t mode) {
 }
 
 uint64_t fmadd_d(uint64_t v1, uint64_t v2, uint64_t v3, uint32_t op, uint8_t mode) {
-    // op should be {softfloat_mulAdd_subProd(2), softfloat_mulAdd_subC(1)}
+    uint64_t F64_SIGN = 1ULL << 63;
+    switch(op) {
+    case 0: // FMADD_D
+        break;
+    case 1: // FMSUB_D
+        v3 ^= F64_SIGN;
+        break;
+    case 2: // FNMADD_D
+        v1 ^= F64_SIGN;
+        v3 ^= F64_SIGN;
+        break;
+    case 3: // FNMSUB_D
+        v1 ^= F64_SIGN;
+        break;
+    }
     softfloat_roundingMode = rmm_map.at(mode);
     softfloat_exceptionFlags = 0;
-    float64_t res = softfloat_mulAddF64(v1, v2, v3, op & 0x1);
-    if(op > 1)
-        res.v ^= 1ULL << 63;
+    float64_t res = softfloat_mulAddF64(v1, v2, v3, 0);
     return res.v;
 }
 
@@ -377,8 +393,8 @@ uint64_t fclass_d(uint64_t v1) {
     uA.f = a;
     uiA = uA.ui;
 
-    uint_fast16_t infOrNaN = expF64UI(uiA) == 0x7FF;
-    uint_fast16_t subnormalOrZero = expF64UI(uiA) == 0;
+    bool infOrNaN = expF64UI(uiA) == 0x7FF;
+    bool subnormalOrZero = expF64UI(uiA) == 0;
     bool sign = signF64UI(uiA);
     bool fracZero = fracF64UI(uiA) == 0;
     bool isNaN = isNaNF64UI(uiA);

src/vm/interp/vm_tgc5c.cpp

@@ -96,7 +96,8 @@ protected:
     using compile_ret_t = virt_addr_t;
     using compile_func = compile_ret_t (this_class::*)(virt_addr_t &pc, code_word_t instr);
 
-    inline const char *name(size_t index){return index<traits::reg_aliases.size()?traits::reg_aliases[index]:"illegal";}
+    inline const char *name(size_t index){return traits::reg_aliases.at(index);}
+
 
     virt_addr_t execute_inst(finish_cond_e cond, virt_addr_t start, uint64_t icount_limit) override;
 
@@ -274,9 +275,6 @@ template <typename CODE_WORD> void debug_fn(CODE_WORD insn) {
     volatile CODE_WORD x = insn;
     insn = 2 * x;
 }
-
-template <typename ARCH> vm_impl<ARCH>::vm_impl() { this(new ARCH()); }
-
 // according to
 // https://stackoverflow.com/questions/8871204/count-number-of-1s-in-binary-representation
 #ifdef __GCC__
@@ -332,17 +330,21 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
     while(!this->core.should_stop() &&
             !(is_icount_limit_enabled(cond) && icount >= count_limit) &&
             !(is_fcount_limit_enabled(cond) && fetch_count >= count_limit)){
-        fetch_count++;
+        if(this->debugging_enabled())
+            this->tgt_adapter->check_continue(*PC);
+        pc.val=*PC;
         if(fetch_ins(pc, data)!=iss::Ok){
-            this->do_sync(POST_SYNC, std::numeric_limits<unsigned>::max());
-            pc.val = super::core.enter_trap(std::numeric_limits<uint64_t>::max(), pc.val, 0);
+            if(this->sync_exec && PRE_SYNC) this->do_sync(PRE_SYNC, std::numeric_limits<unsigned>::max());
+            process_spawn_blocks();
+            if(this->sync_exec && POST_SYNC) this->do_sync(PRE_SYNC, std::numeric_limits<unsigned>::max());
+            pc.val = super::core.enter_trap(arch::traits<ARCH>::RV_CAUSE_FETCH_ACCESS<<16, pc.val, 0);
         } else {
             if (is_jump_to_self_enabled(cond) &&
                     (instr == 0x0000006f || (instr&0xffff)==0xa001)) throw simulation_stopped(0); // 'J 0' or 'C.J 0'
             uint32_t inst_index = instr_decoder.decode_instr(instr);
             opcode_e inst_id = arch::traits<ARCH>::opcode_e::MAX_OPCODE;;
             if(inst_index <instr_descr.size())
-                inst_id = instr_descr.at(instr_decoder.decode_instr(instr)).op;
+                inst_id = instr_descr[inst_index].op;
 
             // pre execution stuff
             this->core.reg.last_branch = 0;
@@ -704,9 +706,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                     }
                                     else {
                                         uint32_t load_address = (uint32_t)((uint64_t)(*(X+rs1) ) + (uint64_t)((int16_t)sext<12>(imm) ));
-                                        int8_t res_27 = super::template read_mem<int8_t>(traits::MEM, load_address);
+                                        int8_t res_1 = super::template read_mem<int8_t>(traits::MEM, load_address);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                        int8_t res = (int8_t)res_27;
+                                        int8_t res = (int8_t)res_1;
                                         if(rd != 0) {
                                             *(X+rd) = (uint32_t)res;
                                         }
@@ -735,9 +737,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                     }
                                     else {
                                         uint32_t load_address = (uint32_t)((uint64_t)(*(X+rs1) ) + (uint64_t)((int16_t)sext<12>(imm) ));
-                                        int16_t res_28 = super::template read_mem<int16_t>(traits::MEM, load_address);
+                                        int16_t res_2 = super::template read_mem<int16_t>(traits::MEM, load_address);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                        int16_t res = (int16_t)res_28;
+                                        int16_t res = (int16_t)res_2;
                                         if(rd != 0) {
                                             *(X+rd) = (uint32_t)res;
                                         }
@@ -766,9 +768,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                     }
                                     else {
                                         uint32_t load_address = (uint32_t)((uint64_t)(*(X+rs1) ) + (uint64_t)((int16_t)sext<12>(imm) ));
-                                        int32_t res_29 = super::template read_mem<int32_t>(traits::MEM, load_address);
+                                        int32_t res_3 = super::template read_mem<int32_t>(traits::MEM, load_address);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                        int32_t res = (int32_t)res_29;
+                                        int32_t res = (int32_t)res_3;
                                         if(rd != 0) {
                                             *(X+rd) = (uint32_t)res;
                                         }
@@ -797,9 +799,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                     }
                                     else {
                                         uint32_t load_address = (uint32_t)((uint64_t)(*(X+rs1) ) + (uint64_t)((int16_t)sext<12>(imm) ));
-                                        uint8_t res_30 = super::template read_mem<uint8_t>(traits::MEM, load_address);
+                                        uint8_t res_4 = super::template read_mem<uint8_t>(traits::MEM, load_address);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                        uint8_t res = res_30;
+                                        uint8_t res = res_4;
                                         if(rd != 0) {
                                             *(X+rd) = (uint32_t)res;
                                         }
@@ -828,9 +830,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                     }
                                     else {
                                         uint32_t load_address = (uint32_t)((uint64_t)(*(X+rs1) ) + (uint64_t)((int16_t)sext<12>(imm) ));
-                                        uint16_t res_31 = super::template read_mem<uint16_t>(traits::MEM, load_address);
+                                        uint16_t res_5 = super::template read_mem<uint16_t>(traits::MEM, load_address);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                        uint16_t res = res_31;
+                                        uint16_t res = res_5;
                                         if(rd != 0) {
                                             *(X+rd) = (uint32_t)res;
                                         }
@@ -1457,7 +1459,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                 case arch::traits<ARCH>::opcode_e::ECALL: {
                     if(this->disass_enabled){
                         /* generate console output when executing the command */
-                        this->core.disass_output(pc.val, "ecall");
+                        //No disass specified, using instruction name
+                        std::string mnemonic = "ecall";
+                        this->core.disass_output(pc.val, mnemonic);
                     }
                     // used registers// calculate next pc value
                     *NEXT_PC = *PC + 4;
@@ -1470,7 +1474,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                 case arch::traits<ARCH>::opcode_e::EBREAK: {
                     if(this->disass_enabled){
                         /* generate console output when executing the command */
-                        this->core.disass_output(pc.val, "ebreak");
+                        //No disass specified, using instruction name
+                        std::string mnemonic = "ebreak";
+                        this->core.disass_output(pc.val, mnemonic);
                     }
                     // used registers// calculate next pc value
                     *NEXT_PC = *PC + 4;
@@ -1483,7 +1489,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                 case arch::traits<ARCH>::opcode_e::MRET: {
                     if(this->disass_enabled){
                         /* generate console output when executing the command */
-                        this->core.disass_output(pc.val, "mret");
+                        //No disass specified, using instruction name
+                        std::string mnemonic = "mret";
+                        this->core.disass_output(pc.val, mnemonic);
                     }
                     // used registers// calculate next pc value
                     *NEXT_PC = *PC + 4;
@@ -1496,7 +1504,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                 case arch::traits<ARCH>::opcode_e::WFI: {
                     if(this->disass_enabled){
                         /* generate console output when executing the command */
-                        this->core.disass_output(pc.val, "wfi");
+                        //No disass specified, using instruction name
+                        std::string mnemonic = "wfi";
+                        this->core.disass_output(pc.val, mnemonic);
                     }
                     // used registers// calculate next pc value
                     *NEXT_PC = *PC + 4;
@@ -1528,9 +1538,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                     else {
                                         uint32_t xrs1 = *(X+rs1);
                                         if(rd != 0) {
-                                            uint32_t res_32 = super::template read_mem<uint32_t>(traits::CSR, csr);
+                                            uint32_t res_6 = super::template read_mem<uint32_t>(traits::CSR, csr);
                                             if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                            uint32_t xrd = res_32;
+                                            uint32_t xrd = res_6;
                                             super::template write_mem<uint32_t>(traits::CSR, csr, xrs1);
                                             if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
                                             *(X+rd) = xrd;
@@ -1563,9 +1573,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                         raise(0, traits::RV_CAUSE_ILLEGAL_INSTRUCTION);
                                     }
                                     else {
-                                        uint32_t res_33 = super::template read_mem<uint32_t>(traits::CSR, csr);
+                                        uint32_t res_7 = super::template read_mem<uint32_t>(traits::CSR, csr);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                        uint32_t xrd = res_33;
+                                        uint32_t xrd = res_7;
                                         uint32_t xrs1 = *(X+rs1);
                                         if(rs1 != 0) {
                                             super::template write_mem<uint32_t>(traits::CSR, csr, xrd | xrs1);
@@ -1598,9 +1608,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                         raise(0, traits::RV_CAUSE_ILLEGAL_INSTRUCTION);
                                     }
                                     else {
-                                        uint32_t res_34 = super::template read_mem<uint32_t>(traits::CSR, csr);
+                                        uint32_t res_8 = super::template read_mem<uint32_t>(traits::CSR, csr);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                        uint32_t xrd = res_34;
+                                        uint32_t xrd = res_8;
                                         uint32_t xrs1 = *(X+rs1);
                                         if(rs1 != 0) {
                                             super::template write_mem<uint32_t>(traits::CSR, csr, xrd & ~ xrs1);
@@ -1633,9 +1643,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                         raise(0, traits::RV_CAUSE_ILLEGAL_INSTRUCTION);
                                     }
                                     else {
-                                        uint32_t res_35 = super::template read_mem<uint32_t>(traits::CSR, csr);
+                                        uint32_t res_9 = super::template read_mem<uint32_t>(traits::CSR, csr);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                        uint32_t xrd = res_35;
+                                        uint32_t xrd = res_9;
                                         super::template write_mem<uint32_t>(traits::CSR, csr, (uint32_t)zimm);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
                                         if(rd != 0) {
@@ -1665,9 +1675,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                         raise(0, traits::RV_CAUSE_ILLEGAL_INSTRUCTION);
                                     }
                                     else {
-                                        uint32_t res_36 = super::template read_mem<uint32_t>(traits::CSR, csr);
+                                        uint32_t res_10 = super::template read_mem<uint32_t>(traits::CSR, csr);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                        uint32_t xrd = res_36;
+                                        uint32_t xrd = res_10;
                                         if(zimm != 0) {
                                             super::template write_mem<uint32_t>(traits::CSR, csr, xrd | (uint32_t)zimm);
                                             if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
@@ -1699,9 +1709,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                                         raise(0, traits::RV_CAUSE_ILLEGAL_INSTRUCTION);
                                     }
                                     else {
-                                        uint32_t res_37 = super::template read_mem<uint32_t>(traits::CSR, csr);
+                                        uint32_t res_11 = super::template read_mem<uint32_t>(traits::CSR, csr);
                                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                                        uint32_t xrd = res_37;
+                                        uint32_t xrd = res_11;
                                         if(zimm != 0) {
                                             super::template write_mem<uint32_t>(traits::CSR, csr, xrd & ~ ((uint32_t)zimm));
                                             if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
@@ -1720,7 +1730,7 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                     if(this->disass_enabled){
                         /* generate console output when executing the command */
                         auto mnemonic = fmt::format(
-                            "{mnemonic:10} {rs1}, {rd}, {imm}", fmt::arg("mnemonic", "fence.i"),
+                            "{mnemonic:10} {rs1}, {rd}, {imm}", fmt::arg("mnemonic", "fence_i"),
                             fmt::arg("rs1", name(rs1)), fmt::arg("rd", name(rd)), fmt::arg("imm", imm));
                         this->core.disass_output(pc.val, mnemonic);
                     }
@@ -2036,9 +2046,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                     // execute instruction
                     {
                         uint32_t offs = (uint32_t)((uint64_t)(*(X+rs1 + 8) ) + (uint64_t)(uimm ));
-                        int32_t res_38 = super::template read_mem<int32_t>(traits::MEM, offs);
+                        int32_t res_12 = super::template read_mem<int32_t>(traits::MEM, offs);
                         if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                        *(X+rd + 8) = (uint32_t)(int32_t)res_38;
+                        *(X+rd + 8) = (uint32_t)(int32_t)res_12;
                     }
                     break;
                 }// @suppress("No break at end of case")
@@ -2094,7 +2104,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                     uint8_t nzimm = ((bit_sub<2,5>(instr)) | (bit_sub<12,1>(instr) << 5));
                     if(this->disass_enabled){
                         /* generate console output when executing the command */
-                        this->core.disass_output(pc.val, "c.nop");
+                        //No disass specified, using instruction name
+                        std::string mnemonic = "c.nop";
+                        this->core.disass_output(pc.val, mnemonic);
                     }
                     // used registers// calculate next pc value
                     *NEXT_PC = *PC + 2;
@@ -2200,7 +2212,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                     uint8_t rd = ((bit_sub<7,5>(instr)));
                     if(this->disass_enabled){
                         /* generate console output when executing the command */
-                        this->core.disass_output(pc.val, ".reserved_clui");
+                        //No disass specified, using instruction name
+                        std::string mnemonic = ".reserved_clui";
+                        this->core.disass_output(pc.val, mnemonic);
                     }
                     // used registers// calculate next pc value
                     *NEXT_PC = *PC + 2;
@@ -2458,9 +2472,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                         }
                         else {
                             uint32_t offs = (uint32_t)((uint64_t)(*(X+2) ) + (uint64_t)(uimm ));
-                            int32_t res_39 = super::template read_mem<int32_t>(traits::MEM, offs);
+                            int32_t res_13 = super::template read_mem<int32_t>(traits::MEM, offs);
                             if(this->core.reg.trap_state>=0x80000000UL) throw memory_access_exception();
-                            *(X+rd) = (uint32_t)(int32_t)res_39;
+                            *(X+rd) = (uint32_t)(int32_t)res_13;
                         }
                     }
                     break;
@@ -2519,7 +2533,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                 case arch::traits<ARCH>::opcode_e::__reserved_cmv: {
                     if(this->disass_enabled){
                         /* generate console output when executing the command */
-                        this->core.disass_output(pc.val, ".reserved_cmv");
+                        //No disass specified, using instruction name
+                        std::string mnemonic = ".reserved_cmv";
+                        this->core.disass_output(pc.val, mnemonic);
                     }
                     // used registers// calculate next pc value
                     *NEXT_PC = *PC + 2;
@@ -2585,7 +2601,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                 case arch::traits<ARCH>::opcode_e::C__EBREAK: {
                     if(this->disass_enabled){
                         /* generate console output when executing the command */
-                        this->core.disass_output(pc.val, "c.ebreak");
+                        //No disass specified, using instruction name
+                        std::string mnemonic = "c.ebreak";
+                        this->core.disass_output(pc.val, mnemonic);
                     }
                     // used registers// calculate next pc value
                     *NEXT_PC = *PC + 2;
@@ -2624,7 +2642,9 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                 case arch::traits<ARCH>::opcode_e::DII: {
                     if(this->disass_enabled){
                         /* generate console output when executing the command */
-                        this->core.disass_output(pc.val, "dii");
+                        //No disass specified, using instruction name
+                        std::string mnemonic = "dii";
+                        this->core.disass_output(pc.val, mnemonic);
                     }
                     // used registers// calculate next pc value
                     *NEXT_PC = *PC + 2;
@@ -2654,11 +2674,11 @@ typename vm_base<ARCH>::virt_addr_t vm_impl<ARCH>::execute_inst(finish_cond_e co
                 icount++;
                 instret++;
             }
-            cycle++;
-            pc.val=*NEXT_PC;
-            this->core.reg.PC = this->core.reg.NEXT_PC;
+            *PC = *NEXT_PC;
             this->core.reg.trap_state =  this->core.reg.pending_trap;
         }
+        fetch_count++;
+        cycle++;
     }
     return pc;
 }