added global symbol support!

fibs.asm

@@ -4,6 +4,7 @@ fibs:   .space 56
 
         .text
 main:
+func_fibs:
         addi s1, zero, 0     # storage index
         addi s2, zero, 56    # last storage index
         addi t0, zero, 1     # t0 = F_{i}
@@ -20,3 +21,8 @@ loop:
         addi a7, zero, 93
         ebreak               # launch debugger
         scall                # exit with code 0
+
+
+.set test_val, 0xFF
+
+.global func_fibs

riscemu/Exceptions.py

@@ -44,6 +44,14 @@ def ASSERT_IN(a1, a2):
         raise ParseException("ASSERTION_FAILED: Expected {} to not be in {}".format(a1, a2), (a1, a2))
 
 
+class LinkerException(RiscemuBaseException):
+    def __init__(self, msg, data):
+        self.msg = msg
+        self.data = data
+    def message(self):
+        return FMT_PARSE + "{}(\"{}\", data={})".format(self.__class__.__name__, self.msg, self.data) + FMT_NONE
+
+
 # MMU Exceptions
 
 class MemoryAccessException(RiscemuBaseException):
@@ -64,7 +72,7 @@ class MemoryAccessException(RiscemuBaseException):
         ) + FMT_NONE
 
 
-class OutOfMemoryEsception(RiscemuBaseException):
+class OutOfMemoryException(RiscemuBaseException):
     def __init__(self, action):
         self.action = action
 

riscemu/Executable.py

@@ -66,15 +66,17 @@ class Executable:
     sections: Dict[str, MemorySection]
     symbols: Dict[str, Tuple[str, int]]
     stack_pref: Optional[int]
+    exported_symbols: List[str]
     name: str
 
     def __repr__(self):
-        return "{}(sections = {}, symbols = {}, stack = {}, run_ptr = {})".format(
+        return "{}(sections = {}, symbols = {}, stack = {}, run_ptr = {}, globals={})".format(
             self.__class__.__name__,
             " ".join(self.sections.keys()),
             " ".join(self.symbols.keys()),
             self.stack_pref,
-            self.run_ptr
+            self.run_ptr,
+            ",".join(self.exported_symbols)
         )
 
 
@@ -98,8 +100,8 @@ class LoadedInstruction:
             raise ParseException("Instruction {} expected argument at {} (args: {})".format(self.name, num, self.args))
         arg = self.args[num]
         # look up symbols
-        if arg in self.bin.symbols:
+        if self.bin.has_symb(arg):
-            return self.bin.symbols[arg]
+            return self.bin.lookup_symbol(arg)
         return parse_numeric_argument(arg)
 
     def get_imm_reg(self, num: int):
@@ -111,8 +113,8 @@ class LoadedInstruction:
         arg = self.args[num]
         ASSERT_IN("(", arg)
         imm, reg = arg[:-1].split("(")
-        if imm in self.bin.symbols:
+        if self.bin.has_symb(imm):
-            return self.bin.symbols[imm], reg
+            return self.bin.lookup_symbol(imm), reg
         return parse_numeric_argument(imm), reg
 
     def get_reg(self, num: int):
@@ -244,13 +246,17 @@ class LoadedExecutable:
     symbols: Dict[str, int]
     run_ptr: int
     stack_heap: Tuple[int, int]  # pointers to stack and heap, are nullptr if no stack/heap is available
+    exported_symbols: Dict[str, int]
+    global_symbol_table: Dict[str, int]
 
-    def __init__(self, exe: Executable, base_addr: int):
+    def __init__(self, exe: Executable, base_addr: int, global_symbol_table: Dict[str, int]):
         self.name = exe.name
         self.base_addr = base_addr
         self.sections = list()
         self.sections_by_name = dict()
         self.symbols = dict()
+        self.exported_symbols = dict()
+        self.global_symbol_table = dict()
 
         # stack/heap if wanted
         if exe.stack_pref is not None:
@@ -281,8 +287,14 @@ class LoadedExecutable:
             curr = align_addr(loaded_sec.size + curr)
 
         for name, (sec_name, offset) in exe.symbols.items():
-            ASSERT_IN(sec_name, self.sections_by_name)
+            if sec_name == '_static_':
-            self.symbols[name] = self.sections_by_name[sec_name].base + offset
+                self.symbols[name] = offset
+            else:
+                ASSERT_IN(sec_name, self.sections_by_name)
+                self.symbols[name] = self.sections_by_name[sec_name].base + offset
+
+        for name in exe.exported_symbols:
+            self.exported_symbols[name] = self.symbols[name]
 
         self.size = curr - base_addr
 
@@ -290,6 +302,14 @@ class LoadedExecutable:
         run_ptr_sec, run_ptr_off = exe.run_ptr
         self.run_ptr = self.sections_by_name[run_ptr_sec].base + run_ptr_off
 
+    def lookup_symbol(self, name):
+        if name in self.symbols:
+            return self.symbols[name]
+        if name in self.global_symbol_table:
+            return self.global_symbol_table[name]
+        raise LinkerException('Symbol {} not found!'.format(name), (self,))
+
+
     def __repr__(self):
         return '{}[{}](base=0x{:08X}, size={}bytes, sections={}, run_ptr=0x{:08X})'.format(
             self.__class__.__name__,
@@ -299,3 +319,6 @@ class LoadedExecutable:
             " ".join(self.sections_by_name.keys()),
             self.run_ptr
         )
+
+    def has_symb(self, arg):
+        return arg in self.symbols or arg in self.global_symbol_table

riscemu/ExecutableParser.py

@@ -36,7 +36,7 @@ class ExecutableParser:
             start_ptr = self.symbols['_start']
         elif 'main' in self.symbols:
             start_ptr = self.symbols['main']
-        return Executable(start_ptr, self.sections, self.symbols, self.stack_pref, self.tokenizer.name)
+        return Executable(start_ptr, self.sections, self.symbols, self.stack_pref, self.globals, self.tokenizer.name)
 
     def parse_instruction(self, ins: 'RiscVInstructionToken'):
         if self.active_section is None:
@@ -103,7 +103,7 @@ class ExecutableParser:
 
     def op_global(self, op: 'RiscVPseudoOpToken'):
         ASSERT_LEN(op.args, 1)
-        name = op.args[1]
+        name = op.args[0]
         self.globals.append(name)
 
     def op_set(self, op: 'RiscVPseudoOpToken'):

riscemu/MMU.py

@@ -1,7 +1,7 @@
 from .Config import RunConfig
 from .Executable import Executable, LoadedExecutable, LoadedMemorySection
 from .helpers import align_addr
-from .Exceptions import OutOfMemoryEsception
+from .Exceptions import OutOfMemoryException
 from typing import Dict, List, Tuple, Optional
 
 
@@ -20,6 +20,7 @@ class MMU:
         self.binaries = list()
         self.last_bin = None
         self.conf = conf
+        self.global_symbols = dict()
 
     def load_bin(self, bin: Executable):
         if self.last_bin is None:
@@ -33,10 +34,10 @@ class MMU:
         if bin.stack_pref is None:
             bin.stack_pref = self.conf.preffered_stack_size
 
-        loaded_bin = LoadedExecutable(bin, addr)
+        loaded_bin = LoadedExecutable(bin, addr, self.global_symbols)
 
         if loaded_bin.size + addr > self.max_size:
-            raise OutOfMemoryEsception('load of executable')
+            raise OutOfMemoryException('load of executable')
 
         self.binaries.append(loaded_bin)
         self.last_bin = loaded_bin
@@ -45,6 +46,8 @@ class MMU:
         for sec in loaded_bin.sections:
             self.sections.append(sec)
 
+        self.global_symbols.update(loaded_bin.exported_symbols)
+
         print("Successfully loaded {}".format(loaded_bin))
 
         return loaded_bin

riscemu/Tokenizer.py

@@ -23,7 +23,7 @@ PSEUDO_OPS = [
     '.data',
     '.half',
     '.text',
-    '.word'
+    '.word',
     '.set',
 ]
 

riscemu/__init__.py

@@ -1,4 +1,4 @@
-from .Exceptions import ASSERT_NOT_NULL, ASSERT_LEN, ASSERT_IN, ASSERT_EQ, ASSERT_NOT_IN
+from .Exceptions import *
 
 from .Tokenizer import RiscVToken, RiscVInput, RiscVTokenizer, RiscVInstructionToken, RiscVSymbolToken, \
     RiscVPseudoOpToken, TokenType

riscemu/__main__.py

@@ -2,6 +2,7 @@ if __name__ == '__main__':
     from . import *
     from .helpers import *
     import argparse
+    import sys
 
     parser = argparse.ArgumentParser(description='RISC-V Userspace parser and emulator', prog='riscemu')
     parser.add_argument('file', metavar='file.asm', type=str, help='The assembly file to interpret and run')
@@ -33,15 +34,21 @@ if __name__ == '__main__':
         FMT_NONE = ""
         FMT_PRINT = ""
 
-    tk = RiscVTokenizer(RiscVInput.from_file(args.file))
+    try:
-    tk.tokenize()
+        tk = RiscVTokenizer(RiscVInput.from_file(args.file))
-
+        tk.tokenize()
-    if args.print_tokens:
+
-        print(FMT_PRINT + "Tokens:" + FMT_NONE)
+        if args.print_tokens:
-        for token in tk.tokens:
+            print(FMT_PRINT + "Tokens:" + FMT_NONE)
-            print(token)
+            for token in tk.tokens:
-
+                print(token)
-    executable = ExecutableParser(tk).parse()
+
+        executable = ExecutableParser(tk).parse()
+    except RiscemuBaseException as e:
+        print("Error while parsing: {}".format(e.message()))
+        import traceback
+        traceback.print_exception(type(e), e, e.__traceback__)
+        sys.exit(1)
 
     cpu = CPU(cfg)
     le = cpu.load(executable)