1berry/ruby
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Remove the scraper

Browse Source 
Now that we're using the jit function entry point, we don't need the
scraper.  Thank you for your service, scraper. ❤️
This commit is contained in:
Aaron Patterson 2021-07-15 14:30:43 -07:00 committed by Alan Wu
parent 84a1e04e58
commit 41f405c486
12 changed files with 5 additions and 364 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
.gitignore vendored
View File

@ -228,6 +228,3 @@ lcov*.info
/rb_mjit_header.h /rb_mjit_header.h
/mjit_config.h /mjit_config.h
/include/ruby-*/*/rb_mjit_min_header-*.h /include/ruby-*/*/rb_mjit_min_header-*.h
# YJIT
/yjit_hooks.inc

4
common.mk
View File

@ -1111,8 +1111,6 @@ incs: $(INSNS) {$(VPATH)}node_name.inc {$(VPATH)}known_errors.inc \
insns: $(INSNS) insns: $(INSNS)
yjit_hooks.inc: vm.$(OBJEXT)
id.h: $(tooldir)/generic_erb.rb $(srcdir)/template/id.h.tmpl $(srcdir)/defs/id.def id.h: $(tooldir)/generic_erb.rb $(srcdir)/template/id.h.tmpl $(srcdir)/defs/id.def
$(ECHO) generating $@ $(ECHO) generating $@
$(Q) $(BASERUBY) $(tooldir)/generic_erb.rb --output=$@ \ $(Q) $(BASERUBY) $(tooldir)/generic_erb.rb --output=$@ \
@ -17106,7 +17104,6 @@ yjit_compile.$(OBJEXT): {$(VPATH)}vm_sync.h
yjit_compile.$(OBJEXT): {$(VPATH)}yjit.h yjit_compile.$(OBJEXT): {$(VPATH)}yjit.h
yjit_compile.$(OBJEXT): {$(VPATH)}yjit_asm.h yjit_compile.$(OBJEXT): {$(VPATH)}yjit_asm.h
yjit_compile.$(OBJEXT): {$(VPATH)}yjit_compile.c yjit_compile.$(OBJEXT): {$(VPATH)}yjit_compile.c
yjit_compile.$(OBJEXT): {$(VPATH)}yjit_hooks.inc
yjit_compile.$(OBJEXT): {$(VPATH)}yjit_utils.h yjit_compile.$(OBJEXT): {$(VPATH)}yjit_utils.h
yjit_core.$(OBJEXT): $(CCAN_DIR)/check_type/check_type.h yjit_core.$(OBJEXT): $(CCAN_DIR)/check_type/check_type.h
yjit_core.$(OBJEXT): $(CCAN_DIR)/container_of/container_of.h yjit_core.$(OBJEXT): $(CCAN_DIR)/container_of/container_of.h
@ -17503,7 +17500,6 @@ yjit_iface.$(OBJEXT): {$(VPATH)}yjit.rbinc
yjit_iface.$(OBJEXT): {$(VPATH)}yjit_asm.h yjit_iface.$(OBJEXT): {$(VPATH)}yjit_asm.h
yjit_iface.$(OBJEXT): {$(VPATH)}yjit_codegen.h yjit_iface.$(OBJEXT): {$(VPATH)}yjit_codegen.h
yjit_iface.$(OBJEXT): {$(VPATH)}yjit_core.h yjit_iface.$(OBJEXT): {$(VPATH)}yjit_core.h
yjit_iface.$(OBJEXT): {$(VPATH)}yjit_hooks.inc
yjit_iface.$(OBJEXT): {$(VPATH)}yjit_iface.c yjit_iface.$(OBJEXT): {$(VPATH)}yjit_iface.c
yjit_iface.$(OBJEXT): {$(VPATH)}yjit_iface.h yjit_iface.$(OBJEXT): {$(VPATH)}yjit_iface.h
yjit_utils.$(OBJEXT): {$(VPATH)}yjit_asm.h yjit_utils.$(OBJEXT): {$(VPATH)}yjit_asm.h

4
template/Makefile.in
View File

@ -590,7 +590,7 @@ update-known-errors:
$(IFCHANGE) $(srcdir)/defs/known_errors.def - $(IFCHANGE) $(srcdir)/defs/known_errors.def -
INSNS = opt_sc.inc optinsn.inc optunifs.inc insns.inc insns_info.inc \ INSNS = opt_sc.inc optinsn.inc optunifs.inc insns.inc insns_info.inc \
vmtc.inc vm.inc mjit_compile.inc yjit_hooks.inc vmtc.inc vm.inc mjit_compile.inc
$(INSNS): $(srcdir)/insns.def vm_opts.h \ $(INSNS): $(srcdir)/insns.def vm_opts.h \
$(srcdir)/defs/opt_operand.def $(srcdir)/defs/opt_insn_unif.def \ $(srcdir)/defs/opt_operand.def $(srcdir)/defs/opt_insn_unif.def \
@ -610,8 +610,6 @@ $(INSNS): $(srcdir)/insns.def vm_opts.h \
$(tooldir)/ruby_vm/models/instructions_unifications.rb \ $(tooldir)/ruby_vm/models/instructions_unifications.rb \
$(tooldir)/ruby_vm/models/operands_unifications.rb \ $(tooldir)/ruby_vm/models/operands_unifications.rb \
$(tooldir)/ruby_vm/models/trace_instructions.rb \ $(tooldir)/ruby_vm/models/trace_instructions.rb \
$(tooldir)/ruby_vm/models/yjit.rb \
$(tooldir)/ruby_vm/models/yjit/example_instructions.rb \
$(tooldir)/ruby_vm/models/typemap.rb \ $(tooldir)/ruby_vm/models/typemap.rb \
$(tooldir)/ruby_vm/scripts/converter.rb \ $(tooldir)/ruby_vm/scripts/converter.rb \
$(tooldir)/ruby_vm/scripts/insns2vm.rb \ $(tooldir)/ruby_vm/scripts/insns2vm.rb \

4
tool/ruby_vm/models/instructions.rb
View File

@ -13,12 +13,10 @@
require_relative 'bare_instructions' require_relative 'bare_instructions'
require_relative 'operands_unifications' require_relative 'operands_unifications'
require_relative 'instructions_unifications' require_relative 'instructions_unifications'
require_relative 'yjit'
RubyVM::Instructions = RubyVM::BareInstructions.to_a + \ RubyVM::Instructions = RubyVM::BareInstructions.to_a + \
RubyVM::OperandsUnifications.to_a + \ RubyVM::OperandsUnifications.to_a + \
RubyVM::InstructionsUnifications.to_a + \ RubyVM::InstructionsUnifications.to_a
RubyVM::YJIT::ExampleInstructions.to_a
require_relative 'trace_instructions' require_relative 'trace_instructions'

221
tool/ruby_vm/models/yjit.rb
View File

@ -1,221 +0,0 @@
#! /your/favourite/path/to/ruby
# -*- Ruby -*-
# -*- frozen_string_literal: true; -*-
# -*- warn_indent: true; -*-
#
# Copyright (c) 2020 Wu, Alan. All rights reserved.
#
# This file is a part of the programming language Ruby. Permission is hereby
# granted, to either redistribute and/or modify this file, provided that the
# conditions mentioned in the file COPYING are met. Consult the file for
# details.
module RubyVM::YJIT
ScrapeResult = Struct.new(:pre_call_bytes, :post_call_bytes, :disassembly_lines)
class << self
def target_platform
# Note, checking RUBY_PLATRFORM doesn't work when cross compiling
@platform ||= if RUBY_PLATFORM.include?('darwin')
:darwin
elsif RUBY_PLATFORM.include?('linux')
:linux
else
:unknown
end
end
def get_fileoff
# use the load command to figure out the offset to the start of the content of vm.o
`otool -l vm.o`.each_line do |line|
if (fileoff = line[/fileoff (\d+)/, 1])
p [__method__, line] if $DEBUG
return fileoff.to_i
end
end
raise
end
def get_symbol_offset(symbol)
`nm vm.o`.each_line do |line|
if (offset = line[Regexp.compile('(\h+).+' + Regexp.escape(symbol) + '\Z'), 1])
p [__method__, line] if $DEBUG
return Integer(offset, 16)
end
end
raise
end
def readint8b(offset)
bytes = IO.binread('vm.o', 8, offset)
bytes.unpack('q').first # this is native endian but we want little endian. it's fine if the host moachine is x86
end
def get_symbol_section_and_offset(name)
`objdump -w -t vm.o`.each_line do |line|
split_line = line.split
next unless split_line.size >= 6
# the table should go into a data section
if split_line[5].include?('insns_address_table') && split_line[3].include?('data')
p line if $DEBUG
return [split_line[3], Integer(split_line[0], 16)]
end
end
raise 'Failed to find section and offset for the instruction address table'
end
def get_handler_offset(table_section, table_offset, insn_id)
target_offset = insn_id * 8 + table_offset
reloc_start_message = "RELOCATION RECORDS FOR [#{table_section}]:"
`objdump -w -r vm.o`.each_line do |line|
line.strip!
if (line == reloc_start_message)...(line.empty?)
split_line = line.split
next if split_line.first == 'RELOCATION'
next if split_line == ['OFFSET', 'TYPE', 'VALUE']
if Integer(split_line.first, 16) == target_offset
section, offset = split_line[2].split('+')
p line if $DEBUG
return section, Integer(offset, 16)
end
end
end
raise 'Failed to find relocation info for the target instruction'
end
def objdump_disassemble_command(offset)
case target_platform
when :darwin
"objdump --x86-asm-syntax=intel --start-address=#{offset} --stop-address=#{offset+50} -d vm.o"
when :linux
"objdump -M intel --start-address=#{offset} --stop-address=#{offset+50} -d vm.o"
else
raise "unkown platform"
end
end
def disassemble(offset)
command = objdump_disassemble_command(offset)
puts "Running: #{command}"
disassembly = `#{command}`
instructions = []
puts disassembly if $DEBUG
disassembly.each_line do |line|
line = line.strip
match_data = /\s*\h+:\s*((?:\h\h\s)+)\s+(\w+)/.match(line)
if match_data
bytes = match_data[1]
mnemonic = match_data[2]
instructions << [bytes, mnemonic, line]
break if mnemonic == 'jmp'
elsif !instructions.empty?
p line
raise "expected a continuous sequence of disassembly lines"
end
end
jmp_idx = instructions.find_index { |_, mnemonic, _| mnemonic == 'jmp' }
raise 'failed to find jmp' unless jmp_idx
raise 'generated code for example too long' unless jmp_idx < 10
handler_instructions = instructions[(0..jmp_idx)]
disassembly_lines = handler_instructions.map {|_, _, line| line}
puts "Disassembly for the example handler:"
puts disassembly_lines
raise 'rip reference in example makes copying unsafe' if handler_instructions.any? { |_, _, full_line| full_line.downcase.include?('rip') }
acceptable_mnemonics = %w(mov jmp lea call endbr64)
unrecognized = nil
handler_instructions.each { |i| unrecognized = i unless acceptable_mnemonics.include?(i[1]) }
raise "found an unrecognized \"#{unrecognized[1]}\" instruction in the example. List of recognized instructions: #{acceptable_mnemonics.join(', ')}" if unrecognized
raise 'found multiple jmp instructions' if handler_instructions.count { |_, mnemonic, _| mnemonic == 'jmp' } > 1
raise "the jmp instruction seems to be relative which isn't copiable" if instructions[jmp_idx][0].split.size > 4
raise 'no call instructions found' if handler_instructions.count { |_, mnemonic, _| mnemonic == 'call' } == 0
raise 'found multiple call instructions' if handler_instructions.count { |_, mnemonic, _| mnemonic == 'call' } > 1
call_idx = handler_instructions.find_index { |_, mnemonic, _| mnemonic == 'call' }
pre_call_bytes = []
post_call_bytes = []
handler_instructions.take(call_idx).each do |bytes, mnemonic, _|
pre_call_bytes += bytes.split
end
handler_instructions[call_idx + 1, handler_instructions.size].each do |bytes, _, _|
post_call_bytes += bytes.split
end
ScrapeResult.new(
comma_separated_hex_string(pre_call_bytes),
comma_separated_hex_string(post_call_bytes),
disassembly_lines
)
end
def darwin_scrape(instruction_id)
fileoff = get_fileoff
tc_table_offset = get_symbol_offset('vm_exec_core.insns_address_table')
vm_exec_core_offset = get_symbol_offset('vm_exec_core')
p instruction_id if $DEBUG
p fileoff if $DEBUG
p tc_table_offset.to_s(16) if $DEBUG
offset_to_insn_in_tc_table = fileoff + tc_table_offset + 8 * instruction_id
p offset_to_insn_in_tc_table if $DEBUG
offset_to_handler_code_from_vm_exec_core = readint8b(offset_to_insn_in_tc_table)
p offset_to_handler_code_from_vm_exec_core if $DEBUG
disassemble(vm_exec_core_offset + offset_to_handler_code_from_vm_exec_core)
end
def linux_scrape(instruction_id)
table_section, table_offset = get_symbol_section_and_offset('vm_exec_core.insns_address_table')
p [table_section, table_offset] if $DEBUG
handler_section, handler_offset = get_handler_offset(table_section, table_offset, instruction_id)
p [handler_section, handler_offset] if $DEBUG
disassemble(handler_offset)
end
def make_result(success, with_pc)
[success ? 1 : 0,
[
['yjit_with_ec', with_pc],
]
]
end
def scrape_instruction(instruction_id)
raise unless instruction_id.is_a?(Integer)
case target_platform
when :darwin
darwin_scrape(instruction_id)
when :linux
linux_scrape(instruction_id)
else
raise 'Unknown platform. Only Mach-O on macOS and ELF on Linux are supported'
end
end
def scrape
with_ec = scrape_instruction(RubyVM::Instructions.find_index { |insn| insn.name == 'yjit_call_example_with_ec' })
make_result(true, with_ec)
rescue => e
print_warning("scrape failed: #{e.message}")
int3 = '0xcc'
failure_result = ScrapeResult.new(int3, int3, ['int3'])
make_result(false, failure_result)
end
def print_warning(text)
text = "yjit warning: #{text}"
text = "\x1b[1m#{text}\x1b[0m" if STDOUT.tty?
STDOUT.puts(text)
end
def comma_separated_hex_string(nums)
nums.map{ |byte| '0x'+byte}.join(', ')
end
end
end
require_relative 'yjit/example_instructions'

69
tool/ruby_vm/models/yjit/example_instructions.rb
View File

@ -1,69 +0,0 @@
#! /your/favourite/path/to/ruby
# -*- Ruby -*-
# -*- frozen_string_literal: true; -*-
# -*- warn_indent: true; -*-
#
# Copyright (c) 2020 Wu, Alan. All rights reserved.
#
# This file is a part of the programming language Ruby. Permission is hereby
# granted, to either redistribute and/or modify this file, provided that the
# conditions mentioned in the file COPYING are met. Consult the file for
# details.
class RubyVM::YJIT::ExampleInstructions
include RubyVM::CEscape
attr_reader :name
def initialize(name)
@name = name
end
def pretty_name
return sprintf "%s(...)(...)(...)", @name
end
def jump_destination
return @orig.name
end
def bin
return sprintf "BIN(%s)", @name
end
def width
1
end
def operands_info
""
end
def rets
return ['...']
end
def pops
return ['...']
end
def attributes
return []
end
def has_attribute? *;
return false
end
def handles_sp?
false
end
def always_leaf?
false
end
def self.to_a
[new('yjit_call_example_with_ec')]
end
end

14
tool/ruby_vm/views/vm.inc.erb
View File

@ -25,20 +25,6 @@
<%= render 'insn_entry', locals: { insn: insn } -%> <%= render 'insn_entry', locals: { insn: insn } -%>
% end % end
% %
% RubyVM::YJIT::ExampleInstructions.to_a.each do |insn|
INSN_ENTRY(yjit_call_example_with_ec)
{
START_OF_ORIGINAL_INSN(yjit_call_example_with_ec);
#if USE_MACHINE_REGS
// assumes USE_MACHINE_REGS, aka reg_pc setup,
// aka #define SET_PC(x) (reg_cfp->pc = reg_pc = (x))
rb_yjit_empty_func_with_ec(GET_CFP(), ec);
RESTORE_REGS();
#endif
END_INSN(yjit_call_example_with_ec);
}
% end
%
% RubyVM::TraceInstructions.to_a.each do |insn| % RubyVM::TraceInstructions.to_a.each do |insn|
<%= render 'trace_instruction', locals: { insn: insn } -%> <%= render 'trace_instruction', locals: { insn: insn } -%>
% end % end

24
tool/ruby_vm/views/yjit_hooks.inc.erb
View File

@ -1,24 +0,0 @@
/* -*- C -*- */
%# Copyright (c) 2020 Wu, Alan. All rights reserved.
%#
%# This file is a part of the programming language Ruby. Permission is hereby
%# granted, to either redistribute and/or modify this file, provided that the
%# conditions mentioned in the file COPYING are met. Consult the file for
%# details.
<%= render 'copyright' %>
<%= render 'notice', locals: {
this_file: 'contains raw instruction bytes that helps YJIT generate code',
edit: __FILE__,
} -%>
% success, byte_arrays = RubyVM::YJIT.scrape
static const uint8_t yjit_scrape_successful = <%= success %>;
% byte_arrays.each do |(prefix, scrape_result)|
// Disassembly:
% scrape_result.disassembly_lines.each do |line|
// <%= line %>
% end
static const uint8_t <%= prefix %>_pre_call_bytes[] = { <%= scrape_result.pre_call_bytes %> };
static const uint8_t <%= prefix %>_post_call_bytes[] = { <%= scrape_result.post_call_bytes %> };
% end

2
vm_exec.h
View File

@ -81,7 +81,7 @@ error !
RSTRING_PTR(rb_iseq_path(reg_cfp->iseq)), \ RSTRING_PTR(rb_iseq_path(reg_cfp->iseq)), \
rb_iseq_line_no(reg_cfp->iseq, reg_pc - reg_cfp->iseq->body->iseq_encoded)); \ rb_iseq_line_no(reg_cfp->iseq, reg_pc - reg_cfp->iseq->body->iseq_encoded)); \
} \ } \
if (USE_INSNS_COUNTER && BIN(insn) != BIN(yjit_call_example_with_ec)) vm_insns_counter_count_insn(BIN(insn)); if (USE_INSNS_COUNTER) vm_insns_counter_count_insn(BIN(insn));
#define INSN_DISPATCH_SIG(insn) #define INSN_DISPATCH_SIG(insn)

2
win32/Makefile.sub
View File

@ -1347,7 +1347,7 @@ $(MJIT_PRECOMPILED_HEADER): $(MJIT_PRECOMPILED_HEADER_NAME)
$(Q) $(MAKE_LINK) $(MJIT_PRECOMPILED_HEADER_NAME:.pch=.pdb) $(arch_hdrdir)/$(MJIT_PRECOMPILED_HEADER_NAME:.pch=.pdb) $(Q) $(MAKE_LINK) $(MJIT_PRECOMPILED_HEADER_NAME:.pch=.pdb) $(arch_hdrdir)/$(MJIT_PRECOMPILED_HEADER_NAME:.pch=.pdb)
INSNS = opt_sc.inc optinsn.inc optunifs.inc insns.inc insns_info.inc \ INSNS = opt_sc.inc optinsn.inc optunifs.inc insns.inc insns_info.inc \
vmtc.inc vm.inc mjit_compile.inc yjit_hooks.inc vmtc.inc vm.inc mjit_compile.inc
!if [exit > insns_rules.mk] !if [exit > insns_rules.mk]
!else if [for %I in ($(INSNS)) do \ !else if [for %I in ($(INSNS)) do \

19
yjit_iface.c
View File

@ -13,7 +13,6 @@
#include "yjit_iface.h" #include "yjit_iface.h"
#include "yjit_codegen.h" #include "yjit_codegen.h"
#include "yjit_core.h" #include "yjit_core.h"
#include "yjit_hooks.inc"
#include "darray.h" #include "darray.h"
#if HAVE_LIBCAPSTONE #if HAVE_LIBCAPSTONE
@ -48,22 +47,6 @@ static const rb_data_type_t yjit_block_type = {
0, 0, RUBY_TYPED_FREE_IMMEDIATELY 0, 0, RUBY_TYPED_FREE_IMMEDIATELY
}; };
// Write the YJIT entry point pre-call bytes
void
cb_write_pre_call_bytes(codeblock_t* cb)
{
for (size_t i = 0; i < sizeof(yjit_with_ec_pre_call_bytes); ++i)
cb_write_byte(cb, yjit_with_ec_pre_call_bytes[i]);
}
// Write the YJIT exit post-call bytes
void
cb_write_post_call_bytes(codeblock_t* cb)
{
for (size_t i = 0; i < sizeof(yjit_with_ec_post_call_bytes); ++i)
cb_write_byte(cb, yjit_with_ec_post_call_bytes[i]);
}
// Get the PC for a given index in an iseq // Get the PC for a given index in an iseq
VALUE * VALUE *
yjit_iseq_pc_at_idx(const rb_iseq_t *iseq, uint32_t insn_idx) yjit_iseq_pc_at_idx(const rb_iseq_t *iseq, uint32_t insn_idx)
@ -1021,7 +1004,7 @@ outgoing_ids(VALUE self)
void void
rb_yjit_init(struct rb_yjit_options *options) rb_yjit_init(struct rb_yjit_options *options)
{ {
if (!yjit_scrape_successful || !PLATFORM_SUPPORTED_P) { if (!PLATFORM_SUPPORTED_P) {
return; return;
} }

3
yjit_iface.h
View File

@ -96,9 +96,6 @@ extern yjit_comment_array_t yjit_code_comments;
RUBY_EXTERN struct rb_yjit_options rb_yjit_opts; RUBY_EXTERN struct rb_yjit_options rb_yjit_opts;
RUBY_EXTERN struct rb_yjit_runtime_counters yjit_runtime_counters; RUBY_EXTERN struct rb_yjit_runtime_counters yjit_runtime_counters;
void cb_write_pre_call_bytes(codeblock_t* cb);
void cb_write_post_call_bytes(codeblock_t* cb);
void yjit_map_addr2insn(void *code_ptr, int insn); void yjit_map_addr2insn(void *code_ptr, int insn);
VALUE *yjit_iseq_pc_at_idx(const rb_iseq_t *iseq, uint32_t insn_idx); VALUE *yjit_iseq_pc_at_idx(const rb_iseq_t *iseq, uint32_t insn_idx);
int yjit_opcode_at_pc(const rb_iseq_t *iseq, const VALUE *pc); int yjit_opcode_at_pc(const rb_iseq_t *iseq, const VALUE *pc);