Gabe Black has uploaded this change for review. (
https://gem5-review.googlesource.com/c/public/gem5/+/56335 )
Change subject: arch: Encapsulate the ucode asm into a class.
......................................................................
arch: Encapsulate the ucode asm into a class.
Store the assembler state in this class, rather than tacking it onto the
parser. This will also pave the way to splitting the parser into a top
level parser, and an assembly block parser which follows different
rules.
Change-Id: Id4498a00ebc276940bc6a2f140a43460664d47da
---
M src/arch/micro_asm.py
1 file changed, 272 insertions(+), 261 deletions(-)
diff --git a/src/arch/micro_asm.py b/src/arch/micro_asm.py
index da52a61..6e3a3a0 100644
--- a/src/arch/micro_asm.py
+++ b/src/arch/micro_asm.py
@@ -103,14 +103,14 @@
super().__init__(name)
self.params = params
- def handle(self, parser, container):
- microop = parser.microops.get(self.name, None)
+ def handle(self, assembler, container):
+ microop = assembler.microops.get(self.name, None)
if not microop:
raise Exception(f'Unrecongized mnemonic: "{self.name}"')
try:
microop = eval(f'_cls({self.params})',
- {'_cls': microop}, parser.symbols)
+ {'_cls': microop}, assembler.symbols)
except:
print_error(f'Error instantiating microop "{self.name}"')
raise
@@ -122,13 +122,14 @@
super().__init__(name, True)
self.params = params
- def handle(self, parser, container):
+ def handle(self, assembler, container):
directive = container.directives.get(self.name, None)
if not directive:
raise Exception(f'Unrecognized directive: "{self.name}"')
local = {'_dir': directive}
try:
- eval(f'_dir({self.params})', {'_dir': directive},
parser.symbols)
+ eval(f'_dir({self.params})',
+ {'_dir': directive}, assembler.symbols)
except:
print_error(f'Error executing directive "{self.name}"')
raise
@@ -138,7 +139,7 @@
super().__init__(name)
self.is_extern = is_extern
- def handle(self, parser, container):
+ def handle(self, assembler, container):
container.add_label(self)
##########################################################################
@@ -152,306 +153,302 @@
print("*** %s" % message)
print()
-##########################################################################
-#
-# Lexer specification
-#
-##########################################################################
+class MicroAssembler:
+ ######################################################################
+ #
+ # Lexer specification
+ #
+ ######################################################################
-# Error handler. Just call exit. Output formatted to work under
-# Emacs compile-mode. Optional 'print_traceback' arg, if set to True,
-# prints a Python stack backtrace too (can be handy when trying to
-# debug the parser itself).
-def error(lineno, string, print_traceback = False):
- # Print a Python stack backtrace if requested.
- if (print_traceback):
- traceback.print_exc()
- if lineno != 0:
- line_str = "%d:" % lineno
- else:
- line_str = ""
- sys.exit("%s %s" % (line_str, string))
+ # Error handler. Just call exit. Output formatted to work under
+ # Emacs compile-mode. Optional 'print_traceback' arg, if set to True,
+ # prints a Python stack backtrace too (can be handy when trying to
+ # debug the parser itself).
+ def error(self, lineno, string, print_traceback = False):
+ # Print a Python stack backtrace if requested.
+ if (print_traceback):
+ traceback.print_exc()
+ if lineno != 0:
+ line_str = "%d:" % lineno
+ else:
+ line_str = ""
+ sys.exit("%s %s" % (line_str, string))
-reserved = ('DEF', 'MACROOP', 'ROM', 'EXTERN')
+ reserved = ('DEF', 'MACROOP', 'ROM', 'EXTERN')
-tokens = reserved + (
- # identifier
- 'ID',
- # arguments for microops and directives
- 'PARAMS',
+ tokens = reserved + (
+ # identifier
+ 'ID',
+ # arguments for microops and directives
+ 'PARAMS',
- 'LPAREN', 'RPAREN',
- 'LBRACE', 'RBRACE',
- 'COLON', 'SEMI', 'DOT',
- 'NEWLINE'
- )
+ 'LPAREN', 'RPAREN',
+ 'LBRACE', 'RBRACE',
+ 'COLON', 'SEMI', 'DOT',
+ 'NEWLINE'
+ )
-# New lines are ignored at the top level, but they end statements in the
-# assembler
-states = (
- ('asm', 'exclusive'),
- ('params', 'exclusive'),
- ('header', 'exclusive'),
-)
+ # New lines are ignored at the top level, but they end statements in
the
+ # assembler
+ states = (
+ ('asm', 'exclusive'),
+ ('params', 'exclusive'),
+ ('header', 'exclusive'),
+ )
-reserved_map = { }
-for r in reserved:
- reserved_map[r.lower()] = r
+ reserved_map = { }
+ for r in reserved:
+ reserved_map[r.lower()] = r
-# Ignore comments
-def t_ANY_COMMENT(t):
- r'\#[^\n]*(?=\n)'
+ # Ignore comments
+ def t_ANY_COMMENT(self, t):
+ r'\#[^\n]*(?=\n)'
-def t_ANY_MULTILINECOMMENT(t):
- r'/\*([^/]|((?<!\*)/))*\*/'
+ def t_ANY_MULTILINECOMMENT(self, t):
+ r'/\*([^/]|((?<!\*)/))*\*/'
-# A colon marks the end of a label. It should follow an ID which will
-# put the lexer in the "params" state. Seeing the colon will put it back
-# in the "asm" state since it knows it saw a label and not a mnemonic.
-def t_params_COLON(t):
- r':'
- t.lexer.pop_state()
- return t
+ # A colon marks the end of a label. It should follow an ID which will
+ # put the lexer in the "params" state. Seeing the colon will put it
back
+ # in the "asm" state since it knows it saw a label and not a mnemonic.
+ def t_params_COLON(self, t):
+ r':'
+ t.lexer.pop_state()
+ return t
-# Parameters are a string of text which don't contain an unescaped
statement
-# statement terminator, ie a newline or semi colon.
-def t_params_PARAMS(t):
- r'([^\n;\\]|(\\[\n;\\]))+'
- t.lineno += t.value.count('\n')
- unescapeParamsRE = re.compile(r'(\\[\n;\\])')
- def unescapeParams(mo):
- val = mo.group(0)
- return val[1]
- t.value = unescapeParamsRE.sub(unescapeParams, t.value)
- t.lexer.pop_state()
- return t
+ # Parameters are a string of text which don't contain an unescaped
+ # statement statement terminator, ie a newline or semi colon.
+ def t_params_PARAMS(self, t):
+ r'([^\n;\\]|(\\[\n;\\]))+'
+ t.lineno += t.value.count('\n')
+ unescapeParamsRE = re.compile(r'(\\[\n;\\])')
+ def unescapeParams(mo):
+ val = mo.group(0)
+ return val[1]
+ t.value = unescapeParamsRE.sub(unescapeParams, t.value)
+ t.lexer.pop_state()
+ return t
-# An "ID" in the micro assembler is either a label, directive, or mnemonic
-# If it's either a directive or a mnemonic, it will be optionally followed
by
-# parameters. If it's a label, the following colon will make the lexer stop
-# looking for parameters.
-def t_asm_ID(t):
- r'[A-Za-z_]\w*'
- t.type = reserved_map.get(t.value, 'ID')
- # If the ID is really "extern", we shouldn't start looking for
parameters
- # yet. The real ID, the label itself, is coming up.
- if t.type != 'EXTERN':
- t.lexer.push_state('params')
- return t
+ # An "ID" in the micro assembler is either a label, directive, or
mnemonic
+ # If it's either a directive or a mnemonic, it will be optionally
followed
+ # by parameters. If it's a label, the following colon will make the
lexer
+ # stop looking for parameters.
+ def t_asm_ID(self, t):
+ r'[A-Za-z_]\w*'
+ t.type = self.reserved_map.get(t.value, 'ID')
+ # If the ID is really "extern", we shouldn't start looking for
+ # parameters yet. The real ID, the label itself, is coming up.
+ if t.type != 'EXTERN':
+ t.lexer.push_state('params')
+ return t
-def t_header_ID(t):
- r'[A-Za-z_]\w*'
- return t
+ def t_header_ID(self, t):
+ r'[A-Za-z_]\w*'
+ return t
-# If there is a label and you're -not- in the assembler (which would be
caught
-# above), don't start looking for parameters.
-def t_ANY_ID(t):
- r'[A-Za-z_]\w*'
- t.type = reserved_map.get(t.value, 'ID')
- if t.type == 'MACROOP':
- t.lexer.push_state('asm')
- t.lexer.push_state('header')
- elif t.type == 'ROM':
- t.lexer.push_state('asm')
- t.lexer.push_state('header')
- return t
+ # If there is a label and you're -not- in the assembler (which would be
+ # caught above), don't start looking for parameters.
+ def t_ANY_ID(self, t):
+ r'[A-Za-z_]\w*'
+ t.type = self.reserved_map.get(t.value, 'ID')
+ if t.type == 'MACROOP':
+ t.lexer.push_state('asm')
+ t.lexer.push_state('header')
+ elif t.type == 'ROM':
+ t.lexer.push_state('asm')
+ t.lexer.push_state('header')
+ return t
-# Braces enter and exit micro assembly
-def t_header_LBRACE(t):
- r'\{'
- t.lexer.pop_state()
- return t
+ # Braces enter and exit micro assembly
+ def t_header_LBRACE(self, t):
+ r'\{'
+ t.lexer.pop_state()
+ return t
-def t_asm_RBRACE(t):
- r'\}'
- t.lexer.pop_state()
- return t
+ def t_asm_RBRACE(self, t):
+ r'\}'
+ t.lexer.pop_state()
+ return t
-# In the micro assembler, do line counting but also return a token. The
-# token is needed by the parser to detect the end of a statement.
-def t_asm_NEWLINE(t):
- r'\n+'
- t.lineno += t.value.count('\n')
- return t
+ # In the micro assembler, do line counting but also return a token. The
+ # token is needed by the parser to detect the end of a statement.
+ def t_asm_NEWLINE(self, t):
+ r'\n+'
+ t.lineno += t.value.count('\n')
+ return t
-# A newline or semi colon when looking for params signals that the
statement
-# is over and the lexer should go back to looking for regular assembly.
-def t_params_NEWLINE(t):
- r'\n+'
- t.lineno += t.value.count('\n')
- t.lexer.pop_state()
- return t
+ # A newline or semi colon when looking for params signals that the
+ # statement is over and the lexer should go back to looking for regular
+ # assembly.
+ def t_params_NEWLINE(self, t):
+ r'\n+'
+ t.lineno += t.value.count('\n')
+ t.lexer.pop_state()
+ return t
-def t_params_SEMI(t):
- r';'
- t.lexer.pop_state()
- return t
+ def t_params_SEMI(self, t):
+ r';'
+ t.lexer.pop_state()
+ return t
-# Unless handled specially above, track newlines only for line counting.
-def t_ANY_NEWLINE(t):
- r'\n+'
- t.lineno += t.value.count('\n')
+ # Unless handled specially above, track newlines only for line
counting.
+ def t_ANY_NEWLINE(self, t):
+ r'\n+'
+ t.lineno += t.value.count('\n')
-# Basic regular expressions to pick out simple tokens
-t_ANY_LPAREN = r'\('
-t_ANY_RPAREN = r'\)'
-t_ANY_SEMI = r';'
-t_ANY_DOT = r'\.'
+ # Basic regular expressions to pick out simple tokens
+ t_ANY_LPAREN = r'\('
+ t_ANY_RPAREN = r'\)'
+ t_ANY_SEMI = r';'
+ t_ANY_DOT = r'\.'
-t_ANY_ignore = ' \t\x0c'
+ t_ANY_ignore = ' \t\x0c'
-def t_ANY_error(t):
- error(t.lineno, "illegal character '%s'" % t.value[0])
- t.skip(1)
+ def t_ANY_error(self, t):
+ error(t.lineno, "illegal character '%s'" % t.value[0])
+ t.skip(1)
-##########################################################################
-#
-# Parser specification
-#
-##########################################################################
+ ######################################################################
+ #
+ # Parser specification
+ #
+ ######################################################################
-# Start symbol for a file which may have more than one "object" defined in
it.
-def p_file(t):
- 'file : opt_objects'
+ # A file which may have one or more "object" defined in it.
+ def p_file(self, t):
+ 'file : opt_objects'
-# The objects are optional.
-def p_opt_objects(t):
- '''opt_objects : objects
- |'''
+ # The objects are optional.
+ def p_opt_objects(self, t):
+ '''opt_objects : objects
+ |'''
-# One or more objects.
-def p_objects(t):
- '''objects : objects object
- | object'''
+ # One or more objects.
+ def p_objects(self, t):
+ '''objects : objects object
+ | object'''
-# Objects can be of various types.
-def p_object(t):
- '''object : rom_block
- | macroop_def'''
+ # Objects can be of various types.
+ def p_object(self, t):
+ '''object : rom_block
+ | macroop_def'''
-# Defines a section of microcode that should go in the current ROM.
-def p_rom_block(t):
- 'rom_block : DEF ROM block SEMI'
- if not t.parser.rom:
- print_error("Rom block found, but no Rom object specified.")
- raise TypeError("Rom block found, but no Rom object was
specified.")
- for statement in t[3].statements:
- statement.handle(t.parser, t.parser.rom)
- t[0] = t.parser.rom
+ # Defines a section of microcode that should go in the current ROM.
+ def p_rom_block(self, t):
+ 'rom_block : DEF ROM block SEMI'
+ if not self.rom:
+ print_error("Rom block found, but no Rom object specified.")
+ raise TypeError("Rom block found, but nowhere to put it.")
+ for statement in t[3].statements:
+ statement.handle(self, self.rom)
+ t[0] = self.rom
-# Defines a macroop that jumps to an external label in the ROM.
-def p_macroop_jump(t):
- 'macroop_def : DEF MACROOP ID LPAREN ID RPAREN SEMI'
- if not t.parser.rom_macroop_type:
- print_error("ROM based macroop found, but no ROM macroop " +
- "class was specified.")
- raise TypeError("ROM based macroop found, but no ROM macroop " +
- "class was specified.")
- macroop = t.parser.rom_macroop_type(t[3], t[5])
- t.parser.macroops[t[3]] = macroop
+ # Defines a macroop that jumps to an external label in the ROM.
+ def p_macroop_jump(self, t):
+ 'macroop_def : DEF MACROOP ID LPAREN ID RPAREN SEMI'
+ if not self.rom_macroop_type:
+ print_error("ROM based macroop found, but no ROM macroop " +
+ "class was specified.")
+ raise TypeError("ROM based macroop found, but no ROM macroop "
+
+ "class was specified.")
+ macroop = self.rom_macroop_type(t[3], t[5])
+ self.macroops[t[3]] = macroop
-# Defines a macroop that is combinationally generated.
-def p_macroop_def(t):
- 'macroop_def : DEF MACROOP ID block SEMI'
- try:
- curop = t.parser.macro_type(t[3])
- except TypeError:
- print_error("Error creating macroop object.")
- raise
- for statement in t[4].statements:
- statement.handle(t.parser, curop)
- t.parser.macroops[t[3]] = curop
+ # Defines a macroop that is combinationally generated.
+ def p_macroop_def(self, t):
+ 'macroop_def : DEF MACROOP ID block SEMI'
+ try:
+ curop = self.macro_type(t[3])
+ except TypeError:
+ print_error("Error creating macroop object.")
+ raise
+ for statement in t[4].statements:
+ statement.handle(self, curop)
+ self.macroops[t[3]] = curop
-# A block of statements
-def p_block(t):
- 'block : LBRACE statements RBRACE'
- block = Block()
- block.statements = t[2]
- t[0] = block
+ # A block of statements
+ def p_block(self, t):
+ 'block : LBRACE statements RBRACE'
+ block = Block()
+ block.statements = t[2]
+ t[0] = block
-# One or more statements
-def p_statements_0(t):
- 'statements : statement'
- t[0] = [t[1]] if t[1] else []
+ # One or more statements
+ def p_statements_0(self, t):
+ 'statements : statement'
+ t[0] = [t[1]] if t[1] else []
-def p_statements_1(t):
- 'statements : statements statement'
- if t[2]:
- t[1].append(t[2])
- t[0] = t[1]
+ def p_statements_1(self, t):
+ 'statements : statements statement'
+ if t[2]:
+ t[1].append(t[2])
+ t[0] = t[1]
-# A statement can be of various types.
-def p_statement(t):
- '''statement : empty_statement
- | label
- | microop
- | directive'''
- t[0] = t[1]
+ # A statement can be of various types.
+ def p_statement(self, t):
+ '''statement : empty_statement
+ | label
+ | microop
+ | directive'''
+ t[0] = t[1]
-# Parameters are optional.
-def p_opt_params_0(t):
- 'opt_params : PARAMS'
- t[0] = t[1]
+ # Parameters are optional.
+ def p_opt_params_0(self, t):
+ 'opt_params : PARAMS'
+ t[0] = t[1]
-def p_opt_params_1(t):
- 'opt_params :'
- t[0] = ""
+ def p_opt_params_1(self, t):
+ 'opt_params :'
+ t[0] = ""
-# Statements are often ended by newlines or a semi colon.
-def p_end_of_statement(t):
- '''end_of_statement : NEWLINE
- | SEMI'''
- pass
+ # Statements are often ended by newlines or a semi colon.
+ def p_end_of_statement(self, t):
+ '''end_of_statement : NEWLINE
+ | SEMI'''
+ pass
-# Ignore empty statements.
-def p_empty_statement(t):
- 'empty_statement : end_of_statement'
- pass
+ # Ignore empty statements.
+ def p_empty_statement(self, t):
+ 'empty_statement : end_of_statement'
+ pass
-# A label in the microcode.
-def p_label_0(t):
- 'label : EXTERN ID COLON'
- t[0] = Label(t[2], True)
+ # A label in the microcode.
+ def p_label_0(self, t):
+ 'label : EXTERN ID COLON'
+ t[0] = Label(t[2], True)
-def p_label_1(t):
- 'label : ID COLON'
- t[0] = Label(t[1], False)
+ def p_label_1(self, t):
+ 'label : ID COLON'
+ t[0] = Label(t[1], False)
-# A microop with optional parameters.
-def p_microop(t):
- 'microop : ID opt_params end_of_statement'
- t[0] = Microop(t[1], t[2])
+ # A microop with optional parameters.
+ def p_microop(self, t):
+ 'microop : ID opt_params end_of_statement'
+ t[0] = Microop(t[1], t[2])
-# Directives for the macroop.
-def p_directive(t):
- 'directive : DOT ID opt_params end_of_statement'
- t[0] = Directive(t[2], t[3])
+ # Directives for the macroop.
+ def p_directive(self, t):
+ 'directive : DOT ID opt_params end_of_statement'
+ t[0] = Directive(t[2], t[3])
-# Parse error handler. Note that the argument here is the offending
-# *token*, not a grammar symbol (hence the need to use t.value)
-def p_error(t):
- if t:
- error(t.lineno, "syntax error at '%s'" % t.value)
- else:
- error(0, "unknown syntax error", True)
+ # Parse error handler. Note that the argument here is the offending
+ # *token*, not a grammar symbol (hence the need to use t.value)
+ def p_error(self, t):
+ if t:
+ error(t.lineno, "syntax error at '%s'" % t.value)
+ else:
+ error(0, "unknown syntax error", True)
-class MicroAssembler(object):
-
- def __init__(self, macro_type, microops,
- rom = None, rom_macroop_type = None):
- self.lexer = lex.lex()
- self.parser = yacc.yacc()
- self.parser.macro_type = macro_type
- self.parser.macroops = {}
- self.parser.microops = microops
- self.parser.rom = rom
- self.parser.rom_macroop_type = rom_macroop_type
- self.parser.symbols = {}
- self.symbols = self.parser.symbols
+ def __init__(self, macro_type, microops, rom=None,
rom_macroop_type=None):
+ self.lexer = lex.lex(object=self)
+ self.parser = yacc.yacc(module=self)
+ self.macro_type = macro_type
+ self.macroops = {}
+ self.microops = microops
+ self.rom = rom
+ self.rom_macroop_type = rom_macroop_type
+ self.symbols = {}
def assemble(self, asm):
self.parser.parse(asm, lexer=self.lexer)
- macroops = self.parser.macroops
- self.parser.macroops = {}
- return macroops
+ return self.macroops
--
To view, visit https://gem5-review.googlesource.com/c/public/gem5/+/56335
To unsubscribe, or for help writing mail filters, visit
https://gem5-review.googlesource.com/settings
Gerrit-Project: public/gem5
Gerrit-Branch: develop
Gerrit-Change-Id: Id4498a00ebc276940bc6a2f140a43460664d47da
Gerrit-Change-Number: 56335
Gerrit-PatchSet: 1
Gerrit-Owner: Gabe Black <gabe.bl...@gmail.com>
Gerrit-MessageType: newchange
_______________________________________________
gem5-dev mailing list -- gem5-dev@gem5.org
To unsubscribe send an email to gem5-dev-le...@gem5.org
%(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
