|
|
|
# Copyright 2012-2019 The Meson development team
|
|
|
|
|
|
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
# you may not use this file except in compliance with the License.
|
|
|
|
# You may obtain a copy of the License at
|
|
|
|
|
|
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
|
|
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
# See the License for the specific language governing permissions and
|
|
|
|
# limitations under the License.
|
|
|
|
|
|
|
|
import contextlib, os.path, re, tempfile
|
|
|
|
import collections.abc
|
|
|
|
import itertools
|
|
|
|
import typing as T
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
from functools import lru_cache
|
|
|
|
|
|
|
|
from ..linkers import StaticLinker, GnuLikeDynamicLinkerMixin, SolarisDynamicLinker
|
|
|
|
from .. import coredata
|
|
|
|
from .. import mlog
|
|
|
|
from .. import mesonlib
|
|
|
|
from ..mesonlib import (
|
|
|
|
EnvironmentException, MachineChoice, MesonException,
|
|
|
|
Popen_safe, split_args
|
|
|
|
)
|
|
|
|
from ..envconfig import (
|
|
|
|
Properties,
|
|
|
|
)
|
|
|
|
|
|
|
|
if T.TYPE_CHECKING:
|
|
|
|
from ..coredata import OptionDictType
|
|
|
|
from ..envconfig import MachineInfo
|
|
|
|
from ..environment import Environment
|
|
|
|
from ..linkers import DynamicLinker # noqa: F401
|
|
|
|
|
|
|
|
CompilerType = T.TypeVar('CompilerType', bound=Compiler)
|
|
|
|
|
|
|
|
"""This file contains the data files of all compilers Meson knows
|
|
|
|
about. To support a new compiler, add its information below.
|
|
|
|
Also add corresponding autodetection code in environment.py."""
|
|
|
|
|
|
|
|
header_suffixes = ('h', 'hh', 'hpp', 'hxx', 'H', 'ipp', 'moc', 'vapi', 'di')
|
|
|
|
obj_suffixes = ('o', 'obj', 'res')
|
|
|
|
lib_suffixes = ('a', 'lib', 'dll', 'dll.a', 'dylib', 'so')
|
|
|
|
# Mapping of language to suffixes of files that should always be in that language
|
|
|
|
# This means we can't include .h headers here since they could be C, C++, ObjC, etc.
|
|
|
|
lang_suffixes = {
|
|
|
|
'c': ('c',),
|
|
|
|
'cpp': ('cpp', 'cc', 'cxx', 'c++', 'hh', 'hpp', 'ipp', 'hxx'),
|
|
|
|
'cuda': ('cu',),
|
|
|
|
# f90, f95, f03, f08 are for free-form fortran ('f90' recommended)
|
|
|
|
# f, for, ftn, fpp are for fixed-form fortran ('f' or 'for' recommended)
|
|
|
|
'fortran': ('f90', 'f95', 'f03', 'f08', 'f', 'for', 'ftn', 'fpp'),
|
|
|
|
'd': ('d', 'di'),
|
|
|
|
'objc': ('m',),
|
|
|
|
'objcpp': ('mm',),
|
|
|
|
'rust': ('rs',),
|
|
|
|
'vala': ('vala', 'vapi', 'gs'),
|
|
|
|
'cs': ('cs',),
|
|
|
|
'swift': ('swift',),
|
|
|
|
'java': ('java',),
|
|
|
|
}
|
|
|
|
all_languages = lang_suffixes.keys()
|
|
|
|
cpp_suffixes = lang_suffixes['cpp'] + ('h',)
|
|
|
|
c_suffixes = lang_suffixes['c'] + ('h',)
|
|
|
|
# List of languages that by default consume and output libraries following the
|
|
|
|
# C ABI; these can generally be used interchangebly
|
|
|
|
clib_langs = ('objcpp', 'cpp', 'objc', 'c', 'fortran',)
|
|
|
|
# List of languages that can be linked with C code directly by the linker
|
|
|
|
# used in build.py:process_compilers() and build.py:get_dynamic_linker()
|
|
|
|
clink_langs = ('d', 'cuda') + clib_langs
|
|
|
|
clink_suffixes = ()
|
|
|
|
for _l in clink_langs + ('vala',):
|
|
|
|
clink_suffixes += lang_suffixes[_l]
|
|
|
|
clink_suffixes += ('h', 'll', 's')
|
|
|
|
all_suffixes = set(itertools.chain(*lang_suffixes.values(), clink_suffixes))
|
|
|
|
|
|
|
|
# Languages that should use LDFLAGS arguments when linking.
|
|
|
|
languages_using_ldflags = ('objcpp', 'cpp', 'objc', 'c', 'fortran', 'd', 'cuda')
|
|
|
|
soregex = re.compile(r'.*\.so(\.[0-9]+)?(\.[0-9]+)?(\.[0-9]+)?$')
|
|
|
|
|
|
|
|
# Environment variables that each lang uses.
|
|
|
|
cflags_mapping = {'c': 'CFLAGS',
|
|
|
|
'cpp': 'CXXFLAGS',
|
|
|
|
'cuda': 'CUFLAGS',
|
|
|
|
'objc': 'OBJCFLAGS',
|
|
|
|
'objcpp': 'OBJCXXFLAGS',
|
|
|
|
'fortran': 'FFLAGS',
|
|
|
|
'd': 'DFLAGS',
|
|
|
|
'vala': 'VALAFLAGS',
|
|
|
|
'rust': 'RUSTFLAGS'}
|
|
|
|
|
|
|
|
unixy_compiler_internal_libs = ('m', 'c', 'pthread', 'dl', 'rt')
|
|
|
|
# execinfo is a compiler lib on FreeBSD and NetBSD
|
|
|
|
if mesonlib.is_freebsd() or mesonlib.is_netbsd():
|
|
|
|
unixy_compiler_internal_libs += ('execinfo',)
|
|
|
|
|
|
|
|
# All these are only for C-linkable languages; see `clink_langs` above.
|
|
|
|
|
|
|
|
def sort_clink(lang):
|
|
|
|
'''
|
|
|
|
Sorting function to sort the list of languages according to
|
|
|
|
reversed(compilers.clink_langs) and append the unknown langs in the end.
|
|
|
|
The purpose is to prefer C over C++ for files that can be compiled by
|
|
|
|
both such as assembly, C, etc. Also applies to ObjC, ObjC++, etc.
|
|
|
|
'''
|
|
|
|
if lang not in clink_langs:
|
|
|
|
return 1
|
|
|
|
return -clink_langs.index(lang)
|
|
|
|
|
|
|
|
def is_header(fname):
|
|
|
|
if hasattr(fname, 'fname'):
|
|
|
|
fname = fname.fname
|
|
|
|
suffix = fname.split('.')[-1]
|
|
|
|
return suffix in header_suffixes
|
|
|
|
|
|
|
|
def is_source(fname):
|
|
|
|
if hasattr(fname, 'fname'):
|
|
|
|
fname = fname.fname
|
|
|
|
suffix = fname.split('.')[-1].lower()
|
|
|
|
return suffix in clink_suffixes
|
|
|
|
|
|
|
|
def is_assembly(fname):
|
|
|
|
if hasattr(fname, 'fname'):
|
|
|
|
fname = fname.fname
|
|
|
|
return fname.split('.')[-1].lower() == 's'
|
|
|
|
|
|
|
|
def is_llvm_ir(fname):
|
|
|
|
if hasattr(fname, 'fname'):
|
|
|
|
fname = fname.fname
|
|
|
|
return fname.split('.')[-1] == 'll'
|
|
|
|
|
|
|
|
def is_object(fname):
|
|
|
|
if hasattr(fname, 'fname'):
|
|
|
|
fname = fname.fname
|
|
|
|
suffix = fname.split('.')[-1]
|
|
|
|
return suffix in obj_suffixes
|
|
|
|
|
|
|
|
def is_library(fname):
|
|
|
|
if hasattr(fname, 'fname'):
|
|
|
|
fname = fname.fname
|
|
|
|
|
|
|
|
if soregex.match(fname):
|
|
|
|
return True
|
|
|
|
|
|
|
|
suffix = fname.split('.')[-1]
|
|
|
|
return suffix in lib_suffixes
|
|
|
|
|
|
|
|
def is_known_suffix(fname):
|
|
|
|
if hasattr(fname, 'fname'):
|
|
|
|
fname = fname.fname
|
|
|
|
suffix = fname.split('.')[-1]
|
|
|
|
|
|
|
|
return suffix in all_suffixes
|
|
|
|
|
|
|
|
cuda_buildtype_args = {'plain': [],
|
|
|
|
'debug': [],
|
|
|
|
'debugoptimized': [],
|
|
|
|
'release': [],
|
|
|
|
'minsize': [],
|
|
|
|
}
|
|
|
|
java_buildtype_args = {'plain': [],
|
|
|
|
'debug': ['-g'],
|
|
|
|
'debugoptimized': ['-g'],
|
|
|
|
'release': [],
|
|
|
|
'minsize': [],
|
|
|
|
'custom': [],
|
|
|
|
}
|
|
|
|
|
|
|
|
rust_buildtype_args = {'plain': [],
|
|
|
|
'debug': [],
|
|
|
|
'debugoptimized': [],
|
|
|
|
'release': [],
|
|
|
|
'minsize': [],
|
|
|
|
'custom': [],
|
|
|
|
}
|
|
|
|
|
|
|
|
d_gdc_buildtype_args = {'plain': [],
|
|
|
|
'debug': [],
|
|
|
|
'debugoptimized': ['-finline-functions'],
|
|
|
|
'release': ['-frelease', '-finline-functions'],
|
|
|
|
'minsize': [],
|
|
|
|
'custom': [],
|
|
|
|
}
|
|
|
|
|
|
|
|
d_ldc_buildtype_args = {'plain': [],
|
|
|
|
'debug': [],
|
|
|
|
'debugoptimized': ['-enable-inlining', '-Hkeep-all-bodies'],
|
|
|
|
'release': ['-release', '-enable-inlining', '-Hkeep-all-bodies'],
|
|
|
|
'minsize': [],
|
|
|
|
'custom': [],
|
|
|
|
}
|
|
|
|
|
|
|
|
d_dmd_buildtype_args = {'plain': [],
|
|
|
|
'debug': [],
|
|
|
|
'debugoptimized': ['-inline'],
|
|
|
|
'release': ['-release', '-inline'],
|
|
|
|
'minsize': [],
|
|
|
|
'custom': [],
|
|
|
|
}
|
|
|
|
|
|
|
|
mono_buildtype_args = {'plain': [],
|
|
|
|
'debug': [],
|
|
|
|
'debugoptimized': ['-optimize+'],
|
|
|
|
'release': ['-optimize+'],
|
|
|
|
'minsize': [],
|
|
|
|
'custom': [],
|
|
|
|
}
|
|
|
|
|
|
|
|
swift_buildtype_args = {'plain': [],
|
|
|
|
'debug': [],
|
|
|
|
'debugoptimized': [],
|
|
|
|
'release': [],
|
|
|
|
'minsize': [],
|
|
|
|
'custom': [],
|
|
|
|
}
|
|
|
|
|
|
|
|
gnu_winlibs = ['-lkernel32', '-luser32', '-lgdi32', '-lwinspool', '-lshell32',
|
|
|
|
'-lole32', '-loleaut32', '-luuid', '-lcomdlg32', '-ladvapi32']
|
|
|
|
|
|
|
|
msvc_winlibs = ['kernel32.lib', 'user32.lib', 'gdi32.lib',
|
|
|
|
'winspool.lib', 'shell32.lib', 'ole32.lib', 'oleaut32.lib',
|
|
|
|
'uuid.lib', 'comdlg32.lib', 'advapi32.lib']
|
|
|
|
|
|
|
|
clike_optimization_args = {'0': [],
|
|
|
|
'g': [],
|
|
|
|
'1': ['-O1'],
|
|
|
|
'2': ['-O2'],
|
|
|
|
'3': ['-O3'],
|
|
|
|
's': ['-Os'],
|
|
|
|
}
|
|
|
|
|
|
|
|
cuda_optimization_args = {'0': [],
|
|
|
|
'g': ['-O0'],
|
|
|
|
'1': ['-O1'],
|
|
|
|
'2': ['-O2'],
|
|
|
|
'3': ['-O3'],
|
|
|
|
's': ['-O3']
|
|
|
|
}
|
|
|
|
|
|
|
|
cuda_debug_args = {False: [],
|
|
|
|
True: ['-g']}
|
|
|
|
|
|
|
|
clike_debug_args = {False: [],
|
|
|
|
True: ['-g']}
|
|
|
|
|
|
|
|
base_options = {'b_pch': coredata.UserBooleanOption('Use precompiled headers', True),
|
|
|
|
'b_lto': coredata.UserBooleanOption('Use link time optimization', False),
|
|
|
|
'b_sanitize': coredata.UserComboOption('Code sanitizer to use',
|
|
|
|
['none', 'address', 'thread', 'undefined', 'memory', 'address,undefined'],
|
|
|
|
'none'),
|
|
|
|
'b_lundef': coredata.UserBooleanOption('Use -Wl,--no-undefined when linking', True),
|
|
|
|
'b_asneeded': coredata.UserBooleanOption('Use -Wl,--as-needed when linking', True),
|
|
|
|
'b_pgo': coredata.UserComboOption('Use profile guided optimization',
|
|
|
|
['off', 'generate', 'use'],
|
|
|
|
'off'),
|
|
|
|
'b_coverage': coredata.UserBooleanOption('Enable coverage tracking.',
|
|
|
|
False),
|
|
|
|
'b_colorout': coredata.UserComboOption('Use colored output',
|
|
|
|
['auto', 'always', 'never'],
|
|
|
|
'always'),
|
|
|
|
'b_ndebug': coredata.UserComboOption('Disable asserts',
|
|
|
|
['true', 'false', 'if-release'], 'false'),
|
|
|
|
'b_staticpic': coredata.UserBooleanOption('Build static libraries as position independent',
|
|
|
|
True),
|
|
|
|
'b_pie': coredata.UserBooleanOption('Build executables as position independent',
|
|
|
|
False),
|
|
|
|
'b_bitcode': coredata.UserBooleanOption('Generate and embed bitcode (only macOS/iOS/tvOS)',
|
|
|
|
False),
|
|
|
|
'b_vscrt': coredata.UserComboOption('VS run-time library type to use.',
|
|
|
|
['none', 'md', 'mdd', 'mt', 'mtd', 'from_buildtype'],
|
|
|
|
'from_buildtype'),
|
|
|
|
}
|
|
|
|
|
|
|
|
def option_enabled(boptions, options, option):
|
|
|
|
try:
|
|
|
|
if option not in boptions:
|
|
|
|
return False
|
|
|
|
return options[option].value
|
|
|
|
except KeyError:
|
|
|
|
return False
|
|
|
|
|
|
|
|
def get_base_compile_args(options, compiler):
|
|
|
|
args = []
|
|
|
|
try:
|
|
|
|
if options['b_lto'].value:
|
|
|
|
args.extend(compiler.get_lto_compile_args())
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
try:
|
|
|
|
args += compiler.get_colorout_args(options['b_colorout'].value)
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
try:
|
|
|
|
args += compiler.sanitizer_compile_args(options['b_sanitize'].value)
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
try:
|
|
|
|
pgo_val = options['b_pgo'].value
|
|
|
|
if pgo_val == 'generate':
|
|
|
|
args.extend(compiler.get_profile_generate_args())
|
|
|
|
elif pgo_val == 'use':
|
|
|
|
args.extend(compiler.get_profile_use_args())
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
try:
|
|
|
|
if options['b_coverage'].value:
|
|
|
|
args += compiler.get_coverage_args()
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
try:
|
|
|
|
if (options['b_ndebug'].value == 'true' or
|
|
|
|
(options['b_ndebug'].value == 'if-release' and
|
|
|
|
options['buildtype'].value in {'release', 'plain'})):
|
|
|
|
args += ['-DNDEBUG']
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
# This does not need a try...except
|
|
|
|
if option_enabled(compiler.base_options, options, 'b_bitcode'):
|
|
|
|
args.append('-fembed-bitcode')
|
|
|
|
try:
|
|
|
|
crt_val = options['b_vscrt'].value
|
|
|
|
buildtype = options['buildtype'].value
|
|
|
|
try:
|
|
|
|
args += compiler.get_crt_compile_args(crt_val, buildtype)
|
|
|
|
except AttributeError:
|
|
|
|
pass
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
return args
|
|
|
|
|
|
|
|
def get_base_link_args(options, linker, is_shared_module):
|
|
|
|
args = []
|
|
|
|
try:
|
|
|
|
if options['b_lto'].value:
|
|
|
|
args.extend(linker.get_lto_link_args())
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
try:
|
|
|
|
args += linker.sanitizer_link_args(options['b_sanitize'].value)
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
try:
|
|
|
|
pgo_val = options['b_pgo'].value
|
|
|
|
if pgo_val == 'generate':
|
|
|
|
args.extend(linker.get_profile_generate_args())
|
|
|
|
elif pgo_val == 'use':
|
|
|
|
args.extend(linker.get_profile_use_args())
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
try:
|
|
|
|
if options['b_coverage'].value:
|
|
|
|
args += linker.get_coverage_link_args()
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
|
|
|
|
as_needed = option_enabled(linker.base_options, options, 'b_asneeded')
|
|
|
|
bitcode = option_enabled(linker.base_options, options, 'b_bitcode')
|
|
|
|
# Shared modules cannot be built with bitcode_bundle because
|
|
|
|
# -bitcode_bundle is incompatible with -undefined and -bundle
|
|
|
|
if bitcode and not is_shared_module:
|
|
|
|
args.extend(linker.bitcode_args())
|
|
|
|
elif as_needed:
|
|
|
|
# -Wl,-dead_strip_dylibs is incompatible with bitcode
|
|
|
|
args.extend(linker.get_asneeded_args())
|
|
|
|
|
|
|
|
# Apple's ld (the only one that supports bitcode) does not like any
|
|
|
|
# -undefined arguments at all, so don't pass these when using bitcode
|
|
|
|
if not bitcode:
|
|
|
|
if (not is_shared_module and
|
|
|
|
option_enabled(linker.base_options, options, 'b_lundef')):
|
|
|
|
args.extend(linker.no_undefined_link_args())
|
|
|
|
else:
|
|
|
|
args.extend(linker.get_allow_undefined_link_args())
|
|
|
|
|
|
|
|
try:
|
|
|
|
crt_val = options['b_vscrt'].value
|
|
|
|
buildtype = options['buildtype'].value
|
|
|
|
try:
|
|
|
|
args += linker.get_crt_link_args(crt_val, buildtype)
|
|
|
|
except AttributeError:
|
|
|
|
pass
|
|
|
|
except KeyError:
|
|
|
|
pass
|
|
|
|
return args
|
|
|
|
|
|
|
|
|
|
|
|
class CrossNoRunException(MesonException):
|
|
|
|
pass
|
|
|
|
|
|
|
|
class RunResult:
|
|
|
|
def __init__(self, compiled, returncode=999, stdout='UNDEFINED', stderr='UNDEFINED'):
|
|
|
|
self.compiled = compiled
|
|
|
|
self.returncode = returncode
|
|
|
|
self.stdout = stdout
|
|
|
|
self.stderr = stderr
|
|
|
|
|
|
|
|
class CompilerArgs(collections.abc.MutableSequence):
|
|
|
|
'''
|
|
|
|
List-like class that manages a list of compiler arguments. Should be used
|
|
|
|
while constructing compiler arguments from various sources. Can be
|
|
|
|
operated with ordinary lists, so this does not need to be used
|
|
|
|
everywhere.
|
|
|
|
|
|
|
|
All arguments must be inserted and stored in GCC-style (-lfoo, -Idir, etc)
|
|
|
|
and can converted to the native type of each compiler by using the
|
|
|
|
.to_native() method to which you must pass an instance of the compiler or
|
|
|
|
the compiler class.
|
|
|
|
|
|
|
|
New arguments added to this class (either with .append(), .extend(), or +=)
|
|
|
|
are added in a way that ensures that they override previous arguments.
|
|
|
|
For example:
|
|
|
|
|
|
|
|
>>> a = ['-Lfoo', '-lbar']
|
|
|
|
>>> a += ['-Lpho', '-lbaz']
|
|
|
|
>>> print(a)
|
|
|
|
['-Lpho', '-Lfoo', '-lbar', '-lbaz']
|
|
|
|
|
|
|
|
Arguments will also be de-duped if they can be de-duped safely.
|
|
|
|
|
|
|
|
Note that because of all this, this class is not commutative and does not
|
|
|
|
preserve the order of arguments if it is safe to not. For example:
|
|
|
|
>>> ['-Ifoo', '-Ibar'] + ['-Ifez', '-Ibaz', '-Werror']
|
|
|
|
['-Ifez', '-Ibaz', '-Ifoo', '-Ibar', '-Werror']
|
|
|
|
>>> ['-Ifez', '-Ibaz', '-Werror'] + ['-Ifoo', '-Ibar']
|
|
|
|
['-Ifoo', '-Ibar', '-Ifez', '-Ibaz', '-Werror']
|
|
|
|
|
|
|
|
'''
|
|
|
|
# NOTE: currently this class is only for C-like compilers, but it can be
|
|
|
|
# extended to other languages easily. Just move the following to the
|
|
|
|
# compiler class and initialize when self.compiler is set.
|
|
|
|
|
|
|
|
# Arg prefixes that override by prepending instead of appending
|
|
|
|
prepend_prefixes = ('-I', '-L')
|
|
|
|
# Arg prefixes and args that must be de-duped by returning 2
|
|
|
|
dedup2_prefixes = ('-I', '-isystem', '-L', '-D', '-U')
|
|
|
|
dedup2_suffixes = ()
|
|
|
|
dedup2_args = ()
|
|
|
|
# Arg prefixes and args that must be de-duped by returning 1
|
|
|
|
#
|
|
|
|
# NOTE: not thorough. A list of potential corner cases can be found in
|
|
|
|
# https://github.com/mesonbuild/meson/pull/4593#pullrequestreview-182016038
|
|
|
|
dedup1_prefixes = ('-l', '-Wl,-l', '-Wl,--export-dynamic')
|
|
|
|
dedup1_suffixes = ('.lib', '.dll', '.so', '.dylib', '.a')
|
|
|
|
# Match a .so of the form path/to/libfoo.so.0.1.0
|
|
|
|
# Only UNIX shared libraries require this. Others have a fixed extension.
|
|
|
|
dedup1_regex = re.compile(r'([\/\\]|\A)lib.*\.so(\.[0-9]+)?(\.[0-9]+)?(\.[0-9]+)?$')
|
|
|
|
dedup1_args = ('-c', '-S', '-E', '-pipe', '-pthread')
|
|
|
|
# In generate_link() we add external libs without de-dup, but we must
|
|
|
|
# *always* de-dup these because they're special arguments to the linker
|
|
|
|
always_dedup_args = tuple('-l' + lib for lib in unixy_compiler_internal_libs)
|
|
|
|
|
|
|
|
def __init__(self, compiler: T.Union['Compiler', StaticLinker],
|
|
|
|
iterable: T.Optional[T.Iterable[str]] = None):
|
|
|
|
self.compiler = compiler
|
|
|
|
self.__container = list(iterable) if iterable is not None else [] # type: T.List[str]
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
self.__seen_args = set()
|
|
|
|
for arg in self.__container:
|
|
|
|
self.__seen_args.add(arg)
|
|
|
|
|
|
|
|
@T.overload # noqa: F811
|
|
|
|
def __getitem__(self, index: int) -> str: # noqa: F811
|
|
|
|
pass
|
|
|
|
|
|
|
|
@T.overload # noqa: F811
|
|
|
|
def __getitem__(self, index: slice) -> T.List[str]: # noqa: F811
|
|
|
|
pass
|
|
|
|
|
|
|
|
def __getitem__(self, index): # noqa: F811
|
|
|
|
return self.__container[index]
|
|
|
|
|
|
|
|
@T.overload # noqa: F811
|
|
|
|
def __setitem__(self, index: int, value: str) -> None: # noqa: F811
|
|
|
|
pass
|
|
|
|
|
|
|
|
@T.overload # noqa: F811
|
|
|
|
def __setitem__(self, index: slice, value: T.List[str]) -> None: # noqa: F811
|
|
|
|
pass
|
|
|
|
|
|
|
|
def __setitem__(self, index, value) -> None: # noqa: F811
|
|
|
|
self.__container[index] = value
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
for v in value:
|
|
|
|
self.__seen_args.add(v)
|
|
|
|
|
|
|
|
def __delitem__(self, index: T.Union[int, slice]) -> None:
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
value = self.__container[index]
|
|
|
|
del self.__container[index]
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
if value in self.__seen_args and value in self.__container: # this is also honoring that you can have duplicated entries
|
|
|
|
self.__seen_args.remove(value)
|
|
|
|
|
|
|
|
def __len__(self) -> int:
|
|
|
|
return len(self.__container)
|
|
|
|
|
|
|
|
def insert(self, index: int, value: str) -> None:
|
|
|
|
self.__container.insert(index, value)
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
self.__seen_args.add(value)
|
|
|
|
|
|
|
|
def copy(self) -> 'CompilerArgs':
|
|
|
|
return CompilerArgs(self.compiler, self.__container.copy())
|
|
|
|
|
|
|
|
@classmethod
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
@lru_cache(maxsize=None)
|
|
|
|
def _can_dedup(cls, arg):
|
|
|
|
'''
|
|
|
|
Returns whether the argument can be safely de-duped. This is dependent
|
|
|
|
on three things:
|
|
|
|
|
|
|
|
a) Whether an argument can be 'overridden' by a later argument. For
|
|
|
|
example, -DFOO defines FOO and -UFOO undefines FOO. In this case, we
|
|
|
|
can safely remove the previous occurrence and add a new one. The same
|
|
|
|
is true for include paths and library paths with -I and -L. For
|
|
|
|
these we return `2`. See `dedup2_prefixes` and `dedup2_args`.
|
|
|
|
b) Arguments that once specified cannot be undone, such as `-c` or
|
|
|
|
`-pipe`. New instances of these can be completely skipped. For these
|
|
|
|
we return `1`. See `dedup1_prefixes` and `dedup1_args`.
|
|
|
|
c) Whether it matters where or how many times on the command-line
|
|
|
|
a particular argument is present. This can matter for symbol
|
|
|
|
resolution in static or shared libraries, so we cannot de-dup or
|
|
|
|
reorder them. For these we return `0`. This is the default.
|
|
|
|
|
|
|
|
In addition to these, we handle library arguments specially.
|
|
|
|
With GNU ld, we surround library arguments with -Wl,--start/end-group
|
|
|
|
to recursively search for symbols in the libraries. This is not needed
|
|
|
|
with other linkers.
|
|
|
|
'''
|
|
|
|
# A standalone argument must never be deduplicated because it is
|
|
|
|
# defined by what comes _after_ it. Thus dedupping this:
|
|
|
|
# -D FOO -D BAR
|
|
|
|
# would yield either
|
|
|
|
# -D FOO BAR
|
|
|
|
# or
|
|
|
|
# FOO -D BAR
|
|
|
|
# both of which are invalid.
|
|
|
|
if arg in cls.dedup2_prefixes:
|
|
|
|
return 0
|
|
|
|
if arg.startswith('-L='):
|
|
|
|
# DMD and LDC proxy all linker arguments using -L=; in conjunction
|
|
|
|
# with ld64 on macOS this can lead to command line arguments such
|
|
|
|
# as: `-L=-compatibility_version -L=0 -L=current_version -L=0`.
|
|
|
|
# These cannot be combined, ld64 insists they must be passed with
|
|
|
|
# spaces and quoting does not work. if we deduplicate these then
|
|
|
|
# one of the -L=0 arguments will be removed and the version
|
|
|
|
# argument will consume the next argument instead.
|
|
|
|
return 0
|
|
|
|
if arg in cls.dedup2_args or \
|
|
|
|
arg.startswith(cls.dedup2_prefixes) or \
|
|
|
|
arg.endswith(cls.dedup2_suffixes):
|
|
|
|
return 2
|
|
|
|
if arg in cls.dedup1_args or \
|
|
|
|
arg.startswith(cls.dedup1_prefixes) or \
|
|
|
|
arg.endswith(cls.dedup1_suffixes) or \
|
|
|
|
re.search(cls.dedup1_regex, arg):
|
|
|
|
return 1
|
|
|
|
return 0
|
|
|
|
|
|
|
|
@classmethod
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
@lru_cache(maxsize=None)
|
|
|
|
def _should_prepend(cls, arg):
|
|
|
|
if arg.startswith(cls.prepend_prefixes):
|
|
|
|
return True
|
|
|
|
return False
|
|
|
|
|
|
|
|
def to_native(self, copy: bool = False) -> T.List[str]:
|
|
|
|
# Check if we need to add --start/end-group for circular dependencies
|
|
|
|
# between static libraries, and for recursively searching for symbols
|
|
|
|
# needed by static libraries that are provided by object files or
|
|
|
|
# shared libraries.
|
|
|
|
if copy:
|
|
|
|
new = self.copy()
|
|
|
|
else:
|
|
|
|
new = self
|
|
|
|
# This covers all ld.bfd, ld.gold, ld.gold, and xild on Linux, which
|
|
|
|
# all act like (or are) gnu ld
|
|
|
|
# TODO: this could probably be added to the DynamicLinker instead
|
|
|
|
if (isinstance(self.compiler, Compiler) and
|
|
|
|
self.compiler.linker is not None and
|
|
|
|
isinstance(self.compiler.linker, (GnuLikeDynamicLinkerMixin, SolarisDynamicLinker))):
|
|
|
|
group_start = -1
|
|
|
|
group_end = -1
|
|
|
|
is_soname = False
|
|
|
|
for i, each in enumerate(new):
|
|
|
|
if is_soname:
|
|
|
|
is_soname = False
|
|
|
|
continue
|
|
|
|
elif '-soname' in each:
|
|
|
|
# To proxy these arguments with D you need to split the
|
|
|
|
# arguments, thus you get `-L=-soname -L=lib.so` we don't
|
|
|
|
# want to put the lib in a link -roup
|
|
|
|
is_soname = True
|
|
|
|
continue
|
|
|
|
if not each.startswith(('-Wl,-l', '-l')) and not each.endswith('.a') and \
|
|
|
|
not soregex.match(each):
|
|
|
|
continue
|
|
|
|
group_end = i
|
|
|
|
if group_start < 0:
|
|
|
|
# First occurrence of a library
|
|
|
|
group_start = i
|
|
|
|
if group_start >= 0:
|
|
|
|
# Last occurrence of a library
|
|
|
|
new.insert(group_end + 1, '-Wl,--end-group')
|
|
|
|
new.insert(group_start, '-Wl,--start-group')
|
|
|
|
# Remove system/default include paths added with -isystem
|
|
|
|
if hasattr(self.compiler, 'get_default_include_dirs'):
|
|
|
|
default_dirs = self.compiler.get_default_include_dirs()
|
|
|
|
bad_idx_list = [] # type: T.List[int]
|
|
|
|
for i, each in enumerate(new):
|
|
|
|
# Remove the -isystem and the path if the path is a default path
|
|
|
|
if (each == '-isystem' and
|
|
|
|
i < (len(new) - 1) and
|
|
|
|
new[i + 1] in default_dirs):
|
|
|
|
bad_idx_list += [i, i + 1]
|
|
|
|
elif each.startswith('-isystem=') and each[9:] in default_dirs:
|
|
|
|
bad_idx_list += [i]
|
|
|
|
elif each.startswith('-isystem') and each[8:] in default_dirs:
|
|
|
|
bad_idx_list += [i]
|
|
|
|
for i in reversed(bad_idx_list):
|
|
|
|
new.pop(i)
|
|
|
|
return self.compiler.unix_args_to_native(new.__container)
|
|
|
|
|
|
|
|
def append_direct(self, arg: str) -> None:
|
|
|
|
'''
|
|
|
|
Append the specified argument without any reordering or de-dup except
|
|
|
|
for absolute paths to libraries, etc, which can always be de-duped
|
|
|
|
safely.
|
|
|
|
'''
|
|
|
|
if os.path.isabs(arg):
|
|
|
|
self.append(arg)
|
|
|
|
else:
|
|
|
|
self.__container.append(arg)
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
self.__seen_args.add(arg)
|
|
|
|
|
|
|
|
def extend_direct(self, iterable: T.Iterable[str]) -> None:
|
|
|
|
'''
|
|
|
|
Extend using the elements in the specified iterable without any
|
|
|
|
reordering or de-dup except for absolute paths where the order of
|
|
|
|
include search directories is not relevant
|
|
|
|
'''
|
|
|
|
for elem in iterable:
|
|
|
|
self.append_direct(elem)
|
|
|
|
|
|
|
|
def extend_preserving_lflags(self, iterable: T.Iterable[str]) -> None:
|
|
|
|
normal_flags = []
|
|
|
|
lflags = []
|
|
|
|
for i in iterable:
|
|
|
|
if i not in self.always_dedup_args and (i.startswith('-l') or i.startswith('-L')):
|
|
|
|
lflags.append(i)
|
|
|
|
else:
|
|
|
|
normal_flags.append(i)
|
|
|
|
self.extend(normal_flags)
|
|
|
|
self.extend_direct(lflags)
|
|
|
|
|
|
|
|
def __add__(self, args: T.Iterable[str]) -> 'CompilerArgs':
|
|
|
|
new = self.copy()
|
|
|
|
new += args
|
|
|
|
return new
|
|
|
|
|
|
|
|
def __iadd__(self, args: T.Iterable[str]) -> 'CompilerArgs':
|
|
|
|
'''
|
|
|
|
Add two CompilerArgs while taking into account overriding of arguments
|
|
|
|
and while preserving the order of arguments as much as possible
|
|
|
|
'''
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
this_round_added = set() # a dict that contains a value, when the value was added this round
|
|
|
|
pre = [] # type: T.List[str]
|
|
|
|
post = [] # type: T.List[str]
|
|
|
|
if not isinstance(args, collections.abc.Iterable):
|
|
|
|
raise TypeError('can only concatenate Iterable[str] (not "{}") to CompilerArgs'.format(args))
|
|
|
|
for arg in args:
|
|
|
|
# If the argument can be de-duped, do it either by removing the
|
|
|
|
# previous occurrence of it and adding a new one, or not adding the
|
|
|
|
# new occurrence.
|
|
|
|
dedup = self._can_dedup(arg)
|
|
|
|
if dedup == 1:
|
|
|
|
# Argument already exists and adding a new instance is useless
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
if arg in self.__seen_args or arg in pre or arg in post:
|
|
|
|
continue
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
should_prepend = self._should_prepend(arg)
|
|
|
|
if dedup == 2:
|
|
|
|
# Remove all previous occurrences of the arg and add it anew
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
if arg in self.__seen_args and arg not in this_round_added: #if __seen_args contains arg as well as this_round_added, then its not yet part in self.
|
|
|
|
self.remove(arg)
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
if should_prepend:
|
|
|
|
if arg in pre:
|
|
|
|
pre.remove(arg)
|
|
|
|
else:
|
|
|
|
if arg in post:
|
|
|
|
post.remove(arg)
|
|
|
|
if should_prepend:
|
|
|
|
pre.append(arg)
|
|
|
|
else:
|
|
|
|
post.append(arg)
|
CompilerArgs: make lookup faster
in order to deduplicate arguments as much as possible, we need to check
if a argument is already part of the list. Which is quite slow because
we are checking 3 lists for that.
In smaller projects this might be not of interested. However, in efl
things are quite "heavy", alone generating the ninja file took 40 sec..
16 sec. are spent in __iadd__ of CompilerArgs.
What this patch does to speed this all up is:
1. We only check if a element is in post when we know that it must be in
post same for pre.
2. the checks for if we already do contain a specific value are now done
via a dict, and not via the list.
This overall brings the time from 16 sec. spent in __iadd__ to 9 sec. in
__iadd__.
Another possible solution for all this is to have a internal structure
of CompileArgs of trees, so you have the "basic" layer of arguments
where the position does not matter. Additionally, you have two stacks of
lists, the pre stack and the post stack, duplicates can then be checked
when itereting, which would safe another ~4s in terms of efl. However, i
do not have time for this undertaking right now.
5 years ago
|
|
|
self.__seen_args.add(arg)
|
|
|
|
this_round_added.add(arg)
|
|
|
|
# Insert at the beginning
|
|
|
|
self[:0] = pre
|
|
|
|
# Append to the end
|
|
|
|
self.__container += post
|
|
|
|
return self
|
|
|
|
|
|
|
|
def __radd__(self, args: T.Iterable[str]):
|
|
|
|
new = CompilerArgs(self.compiler, args)
|
|
|
|
new += self
|
|
|
|
return new
|
|
|
|
|
|
|
|
def __eq__(self, other: T.Any) -> T.Union[bool, type(NotImplemented)]:
|
|
|
|
# Only allow equality checks against other CompilerArgs and lists instances
|
|
|
|
if isinstance(other, CompilerArgs):
|
|
|
|
return self.compiler == other.compiler and self.__container == other.__container
|
|
|
|
elif isinstance(other, list):
|
|
|
|
return self.__container == other
|
|
|
|
return NotImplemented
|
|
|
|
|
|
|
|
def append(self, arg: str) -> None:
|
|
|
|
self.__iadd__([arg])
|
|
|
|
|
|
|
|
def extend(self, args: T.Iterable[str]) -> None:
|
|
|
|
self.__iadd__(args)
|
|
|
|
|
|
|
|
def __repr__(self) -> str:
|
|
|
|
return 'CompilerArgs({!r}, {!r})'.format(self.compiler, self.__container)
|
|
|
|
|
|
|
|
class Compiler:
|
|
|
|
# Libraries to ignore in find_library() since they are provided by the
|
|
|
|
# compiler or the C library. Currently only used for MSVC.
|
|
|
|
ignore_libs = ()
|
|
|
|
# Libraries that are internal compiler implementations, and must not be
|
|
|
|
# manually searched.
|
|
|
|
internal_libs = ()
|
|
|
|
|
|
|
|
LINKER_PREFIX = None # type: T.Union[None, str, T.List[str]]
|
|
|
|
INVOKES_LINKER = True
|
|
|
|
|
|
|
|
def __init__(self, exelist, version, for_machine: MachineChoice, info: 'MachineInfo',
|
|
|
|
linker: T.Optional['DynamicLinker'] = None, **kwargs):
|
|
|
|
if isinstance(exelist, str):
|
|
|
|
self.exelist = [exelist]
|
|
|
|
elif isinstance(exelist, list):
|
|
|
|
self.exelist = exelist
|
|
|
|
else:
|
|
|
|
raise TypeError('Unknown argument to Compiler')
|
|
|
|
# In case it's been overridden by a child class already
|
|
|
|
if not hasattr(self, 'file_suffixes'):
|
|
|
|
self.file_suffixes = lang_suffixes[self.language]
|
|
|
|
if not hasattr(self, 'can_compile_suffixes'):
|
|
|
|
self.can_compile_suffixes = set(self.file_suffixes)
|
|
|
|
self.default_suffix = self.file_suffixes[0]
|
|
|
|
self.version = version
|
|
|
|
if 'full_version' in kwargs:
|
|
|
|
self.full_version = kwargs['full_version']
|
|
|
|
else:
|
|
|
|
self.full_version = None
|
|
|
|
self.for_machine = for_machine
|
|
|
|
self.base_options = []
|
|
|
|
self.linker = linker
|
|
|
|
self.info = info
|
|
|
|
|
|
|
|
def __repr__(self):
|
|
|
|
repr_str = "<{0}: v{1} `{2}`>"
|
|
|
|
return repr_str.format(self.__class__.__name__, self.version,
|
|
|
|
' '.join(self.exelist))
|
|
|
|
|
|
|
|
@lru_cache(maxsize=None)
|
|
|
|
def can_compile(self, src) -> bool:
|
|
|
|
if hasattr(src, 'fname'):
|
|
|
|
src = src.fname
|
|
|
|
suffix = os.path.splitext(src)[1].lower()
|
|
|
|
if suffix and suffix[1:] in self.can_compile_suffixes:
|
|
|
|
return True
|
|
|
|
return False
|
|
|
|
|
|
|
|
def get_id(self) -> str:
|
|
|
|
return self.id
|
|
|
|
|
|
|
|
def get_linker_id(self) -> str:
|
|
|
|
# There is not guarantee that we have a dynamic linker instance, as
|
|
|
|
# some languages don't have separate linkers and compilers. In those
|
|
|
|
# cases return the compiler id
|
|
|
|
try:
|
|
|
|
return self.linker.id
|
|
|
|
except AttributeError:
|
|
|
|
return self.id
|
|
|
|
|
|
|
|
def get_version_string(self) -> str:
|
|
|
|
details = [self.id, self.version]
|
|
|
|
if self.full_version:
|
|
|
|
details += ['"%s"' % (self.full_version)]
|
|
|
|
return '(%s)' % (' '.join(details))
|
|
|
|
|
|
|
|
def get_language(self) -> str:
|
|
|
|
return self.language
|
|
|
|
|
|
|
|
@classmethod
|
|
|
|
def get_display_language(cls) -> str:
|
|
|
|
return cls.language.capitalize()
|
|
|
|
|
|
|
|
def get_default_suffix(self) -> str:
|
|
|
|
return self.default_suffix
|
|
|
|
|
|
|
|
def get_define(self, dname, prefix, env, extra_args, dependencies) -> T.Tuple[str, bool]:
|
|
|
|
raise EnvironmentException('%s does not support get_define ' % self.get_id())
|
|
|
|
|
|
|
|
def compute_int(self, expression, low, high, guess, prefix, env, extra_args, dependencies) -> int:
|
|
|
|
raise EnvironmentException('%s does not support compute_int ' % self.get_id())
|
|
|
|
|
|
|
|
def compute_parameters_with_absolute_paths(self, parameter_list, build_dir):
|
|
|
|
raise EnvironmentException('%s does not support compute_parameters_with_absolute_paths ' % self.get_id())
|
|
|
|
|
|
|
|
def has_members(self, typename, membernames, prefix, env, *,
|
|
|
|
extra_args=None, dependencies=None) -> T.Tuple[bool, bool]:
|
|
|
|
raise EnvironmentException('%s does not support has_member(s) ' % self.get_id())
|
|
|
|
|
|
|
|
def has_type(self, typename, prefix, env, extra_args, *,
|
|
|
|
dependencies=None) -> T.Tuple[bool, bool]:
|
|
|
|
raise EnvironmentException('%s does not support has_type ' % self.get_id())
|
|
|
|
|
|
|
|
def symbols_have_underscore_prefix(self, env) -> bool:
|
|
|
|
raise EnvironmentException('%s does not support symbols_have_underscore_prefix ' % self.get_id())
|
|
|
|
|
|
|
|
def get_exelist(self):
|
|
|
|
return self.exelist[:]
|
|
|
|
|
|
|
|
def get_linker_exelist(self) -> T.List[str]:
|
|
|
|
return self.linker.get_exelist()
|
|
|
|
|
|
|
|
def get_linker_output_args(self, outputname: str) -> T.List[str]:
|
|
|
|
return self.linker.get_output_args(outputname)
|
|
|
|
|
|
|
|
def get_builtin_define(self, *args, **kwargs):
|
|
|
|
raise EnvironmentException('%s does not support get_builtin_define.' % self.id)
|
|
|
|
|
|
|
|
def has_builtin_define(self, *args, **kwargs):
|
|
|
|
raise EnvironmentException('%s does not support has_builtin_define.' % self.id)
|
|
|
|
|
|
|
|
def get_always_args(self):
|
|
|
|
return []
|
|
|
|
|
|
|
|
def can_linker_accept_rsp(self) -> bool:
|
|
|
|
"""
|
|
|
|
Determines whether the linker can accept arguments using the @rsp syntax.
|
|
|
|
"""
|
|
|
|
return self.linker.get_accepts_rsp()
|
|
|
|
|
|
|
|
def get_linker_always_args(self):
|
|
|
|
return self.linker.get_always_args()
|
|
|
|
|
|
|
|
def get_linker_lib_prefix(self):
|
|
|
|
return self.linker.get_lib_prefix()
|
|
|
|
|
|
|
|
def gen_import_library_args(self, implibname):
|
|
|
|
"""
|
|
|
|
Used only on Windows for libraries that need an import library.
|
|
|
|
This currently means C, C++, Fortran.
|
|
|
|
"""
|
|
|
|
return []
|
|
|
|
|
|
|
|
def get_linker_args_from_envvars(self) -> T.List[str]:
|
|
|
|
return self.linker.get_args_from_envvars()
|
|
|
|
|
|
|
|
def get_options(self) -> T.Dict[str, coredata.UserOption]:
|
|
|
|
return {}
|
|
|
|
|
|
|
|
def get_option_compile_args(self, options):
|
|
|
|
return []
|
|
|
|
|
|
|
|
def get_option_link_args(self, options: 'OptionDictType') -> T.List[str]:
|
|
|
|
return self.linker.get_option_args(options)
|
|
|
|
|
|
|
|
def check_header(self, *args, **kwargs) -> T.Tuple[bool, bool]:
|
|
|
|
raise EnvironmentException('Language %s does not support header checks.' % self.get_display_language())
|
|
|
|
|
|
|
|
def has_header(self, *args, **kwargs) -> T.Tuple[bool, bool]:
|
|
|
|
raise EnvironmentException('Language %s does not support header checks.' % self.get_display_language())
|
|
|
|
|
|
|
|
def has_header_symbol(self, *args, **kwargs) -> T.Tuple[bool, bool]:
|
|
|
|
raise EnvironmentException('Language %s does not support header symbol checks.' % self.get_display_language())
|
|
|
|
|
|
|
|
def compiles(self, *args, **kwargs) -> T.Tuple[bool, bool]:
|
|
|
|
raise EnvironmentException('Language %s does not support compile checks.' % self.get_display_language())
|
|
|
|
|
|
|
|
def links(self, *args, **kwargs) -> T.Tuple[bool, bool]:
|
|
|
|
raise EnvironmentException('Language %s does not support link checks.' % self.get_display_language())
|
|
|
|
|
|
|
|
def run(self, *args, **kwargs) -> RunResult:
|
|
|
|
raise EnvironmentException('Language %s does not support run checks.' % self.get_display_language())
|
|
|
|
|
|
|
|
def sizeof(self, *args, **kwargs) -> int:
|
|
|
|
raise EnvironmentException('Language %s does not support sizeof checks.' % self.get_display_language())
|
|
|
|
|
|
|
|
def alignment(self, *args, **kwargs) -> int:
|
|
|
|
raise EnvironmentException('Language %s does not support alignment checks.' % self.get_display_language())
|
|
|
|
|
|
|
|
def has_function(self, *args, **kwargs) -> T.Tuple[bool, bool]:
|
|
|
|
raise EnvironmentException('Language %s does not support function checks.' % self.get_display_language())
|
|
|
|
|
|
|
|
@classmethod
|
|
|
|
def unix_args_to_native(cls, args):
|
|
|
|
"Always returns a copy that can be independently mutated"
|
|
|
|
return args[:]
|
|
|
|
|
|
|
|
@classmethod
|
|
|
|
def native_args_to_unix(cls, args: T.List[str]) -> T.List[str]:
|
|
|
|
"Always returns a copy that can be independently mutated"
|
|
|
|
return args[:]
|
|
|
|
|
|
|
|
def find_library(self, *args, **kwargs):
|
|
|
|
raise EnvironmentException('Language {} does not support library finding.'.format(self.get_display_language()))
|
|
|
|
|
|
|
|
def get_library_dirs(self, *args, **kwargs):
|
|
|
|
return ()
|
|
|
|
|
|
|
|
def get_program_dirs(self, *args, **kwargs):
|
|
|
|
return []
|
|
|
|
|
|
|
|
def has_multi_arguments(self, args, env) -> T.Tuple[bool, bool]:
|
|
|
|
raise EnvironmentException(
|
|
|
|
'Language {} does not support has_multi_arguments.'.format(
|
|
|
|
self.get_display_language()))
|
|
|
|
|
|
|
|
def has_multi_link_arguments(self, args: T.List[str], env: 'Environment') -> T.Tuple[bool, bool]:
|
|
|
|
return self.linker.has_multi_arguments(args, env)
|
|
|
|
|
|
|
|
def _get_compile_output(self, dirname, mode):
|
|
|
|
# In pre-processor mode, the output is sent to stdout and discarded
|
|
|
|
if mode == 'preprocess':
|
|
|
|
return None
|
|
|
|
# Extension only matters if running results; '.exe' is
|
|
|
|
# guaranteed to be executable on every platform.
|
|
|
|
if mode == 'link':
|
|
|
|
suffix = 'exe'
|
|
|
|
else:
|
|
|
|
suffix = 'obj'
|
|
|
|
return os.path.join(dirname, 'output.' + suffix)
|
|
|
|
|
|
|
|
def get_compiler_args_for_mode(self, mode):
|
|
|
|
args = []
|
|
|
|
args += self.get_always_args()
|
|
|
|
if mode == 'compile':
|
|
|
|
args += self.get_compile_only_args()
|
|
|
|
if mode == 'preprocess':
|
|
|
|
args += self.get_preprocess_only_args()
|
|
|
|
return args
|
|
|
|
|
|
|
|
@contextlib.contextmanager
|
|
|
|
def compile(self, code, extra_args=None, *, mode='link', want_output=False, temp_dir=None):
|
|
|
|
if extra_args is None:
|
|
|
|
extra_args = []
|
|
|
|
try:
|
|
|
|
with tempfile.TemporaryDirectory(dir=temp_dir) as tmpdirname:
|
|
|
|
if isinstance(code, str):
|
|
|
|
srcname = os.path.join(tmpdirname,
|
|
|
|
'testfile.' + self.default_suffix)
|
|
|
|
with open(srcname, 'w') as ofile:
|
|
|
|
ofile.write(code)
|
|
|
|
elif isinstance(code, mesonlib.File):
|
|
|
|
srcname = code.fname
|
|
|
|
|
|
|
|
# Construct the compiler command-line
|
|
|
|
commands = CompilerArgs(self)
|
|
|
|
commands.append(srcname)
|
|
|
|
# Preprocess mode outputs to stdout, so no output args
|
|
|
|
if mode != 'preprocess':
|
|
|
|
output = self._get_compile_output(tmpdirname, mode)
|
|
|
|
commands += self.get_output_args(output)
|
|
|
|
commands.extend(self.get_compiler_args_for_mode(mode))
|
|
|
|
# extra_args must be last because it could contain '/link' to
|
|
|
|
# pass args to VisualStudio's linker. In that case everything
|
|
|
|
# in the command line after '/link' is given to the linker.
|
|
|
|
commands += extra_args
|
|
|
|
# Generate full command-line with the exelist
|
|
|
|
commands = self.get_exelist() + commands.to_native()
|
|
|
|
mlog.debug('Running compile:')
|
|
|
|
mlog.debug('Working directory: ', tmpdirname)
|
|
|
|
mlog.debug('Command line: ', ' '.join(commands), '\n')
|
|
|
|
mlog.debug('Code:\n', code)
|
|
|
|
os_env = os.environ.copy()
|
|
|
|
os_env['LC_ALL'] = 'C'
|
|
|
|
p, p.stdo, p.stde = Popen_safe(commands, cwd=tmpdirname, env=os_env)
|
|
|
|
mlog.debug('Compiler stdout:\n', p.stdo)
|
|
|
|
mlog.debug('Compiler stderr:\n', p.stde)
|
|
|
|
p.commands = commands
|
|
|
|
p.input_name = srcname
|
|
|
|
if want_output:
|
|
|
|
p.output_name = output
|
|
|
|
p.cached = False # Make sure that the cached attribute always exists
|
|
|
|
yield p
|
|
|
|
except OSError:
|
|
|
|
# On Windows antivirus programs and the like hold on to files so
|
|
|
|
# they can't be deleted. There's not much to do in this case. Also,
|
|
|
|
# catch OSError because the directory is then no longer empty.
|
|
|
|
pass
|
|
|
|
|
|
|
|
@contextlib.contextmanager
|
|
|
|
def cached_compile(self, code, cdata: coredata.CoreData, *, extra_args=None, mode: str = 'link', temp_dir=None):
|
|
|
|
assert(isinstance(cdata, coredata.CoreData))
|
|
|
|
|
|
|
|
# Calculate the key
|
|
|
|
textra_args = tuple(extra_args) if extra_args is not None else None
|
|
|
|
key = (tuple(self.exelist), self.version, code, textra_args, mode)
|
|
|
|
|
|
|
|
# Check if not cached
|
|
|
|
if key not in cdata.compiler_check_cache:
|
|
|
|
with self.compile(code, extra_args=extra_args, mode=mode, want_output=False, temp_dir=temp_dir) as p:
|
|
|
|
# Remove all attributes except the following
|
|
|
|
# This way the object can be serialized
|
|
|
|
tokeep = ['args', 'commands', 'input_name', 'output_name',
|
|
|
|
'pid', 'returncode', 'stdo', 'stde', 'text_mode']
|
|
|
|
todel = [x for x in vars(p).keys() if x not in tokeep]
|
|
|
|
for i in todel:
|
|
|
|
delattr(p, i)
|
|
|
|
p.cached = False
|
|
|
|
cdata.compiler_check_cache[key] = p
|
|
|
|
yield p
|
|
|
|
return
|
|
|
|
|
|
|
|
# Return cached
|
|
|
|
p = cdata.compiler_check_cache[key]
|
|
|
|
p.cached = True
|
|
|
|
mlog.debug('Using cached compile:')
|
|
|
|
mlog.debug('Cached command line: ', ' '.join(p.commands), '\n')
|
|
|
|
mlog.debug('Code:\n', code)
|
|
|
|
mlog.debug('Cached compiler stdout:\n', p.stdo)
|
|
|
|
mlog.debug('Cached compiler stderr:\n', p.stde)
|
|
|
|
yield p
|
|
|
|
|
|
|
|
def get_colorout_args(self, colortype):
|
|
|
|
return []
|
|
|
|
|
|
|
|
# Some compilers (msvc) write debug info to a separate file.
|
|
|
|
# These args specify where it should be written.
|
|
|
|
def get_compile_debugfile_args(self, rel_obj, **kwargs):
|
|
|
|
return []
|
|
|
|
|
|
|
|
def get_link_debugfile_name(self, targetfile: str) -> str:
|
|
|
|
return self.linker.get_debugfile_name(targetfile)
|
|
|
|
|
|
|
|
def get_link_debugfile_args(self, targetfile: str) -> T.List[str]:
|
|
|
|
return self.linker.get_debugfile_args(targetfile)
|
|
|
|
|
|
|
|
def get_std_shared_lib_link_args(self) -> T.List[str]:
|
|
|
|
return self.linker.get_std_shared_lib_args()
|
|
|
|
|
|
|
|
def get_std_shared_module_link_args(self, options: 'OptionDictType') -> T.List[str]:
|
|
|
|
return self.linker.get_std_shared_module_args(options)
|
|
|
|
|
|
|
|
def get_link_whole_for(self, args: T.List[str]) -> T.List[str]:
|
|
|
|
return self.linker.get_link_whole_for(args)
|
|
|
|
|
|
|
|
def get_allow_undefined_link_args(self) -> T.List[str]:
|
|
|
|
return self.linker.get_allow_undefined_args()
|
|
|
|
|
|
|
|
def no_undefined_link_args(self) -> T.List[str]:
|
|
|
|
return self.linker.no_undefined_args()
|
|
|
|
|
|
|
|
# Compiler arguments needed to enable the given instruction set.
|
|
|
|
# May be [] meaning nothing needed or None meaning the given set
|
|
|
|
# is not supported.
|
|
|
|
def get_instruction_set_args(self, instruction_set):
|
|
|
|
return None
|
|
|
|
|
|
|
|
def build_rpath_args(self, env: 'Environment', build_dir: str, from_dir: str,
|
|
|
|
rpath_paths: str, build_rpath: str,
|
|
|
|
install_rpath: str) -> T.List[str]:
|
|
|
|
return self.linker.build_rpath_args(
|
|
|
|
env, build_dir, from_dir, rpath_paths, build_rpath, install_rpath)
|
|
|
|
|
|
|
|
def thread_flags(self, env):
|
|
|
|
return []
|
|
|
|
|
|
|
|
def openmp_flags(self):
|
|
|
|
raise EnvironmentException('Language %s does not support OpenMP flags.' % self.get_display_language())
|
|
|
|
|
|
|
|
def language_stdlib_only_link_flags(self):
|
|
|
|
return []
|
|
|
|
|
|
|
|
def gnu_symbol_visibility_args(self, vistype):
|
|
|
|
return []
|
|
|
|
|
|
|
|
def get_gui_app_args(self, value):
|
|
|
|
return []
|
|
|
|
|
|
|
|
def has_func_attribute(self, name, env):
|
|
|
|
raise EnvironmentException(
|
|
|
|
'Language {} does not support function attributes.'.format(self.get_display_language()))
|
|
|
|
|
|
|
|
def get_pic_args(self):
|
|
|
|
m = 'Language {} does not support position-independent code'
|
|
|
|
raise EnvironmentException(m.format(self.get_display_language()))
|
|
|
|
|
|
|
|
def get_pie_args(self):
|
|
|
|
m = 'Language {} does not support position-independent executable'
|
|
|
|
raise EnvironmentException(m.format(self.get_display_language()))
|
|
|
|
|
|
|
|
def get_pie_link_args(self) -> T.List[str]:
|
|
|
|
return self.linker.get_pie_args()
|
|
|
|
|
|
|
|
def get_argument_syntax(self):
|
|
|
|
"""Returns the argument family type.
|
|
|
|
|
|
|
|
Compilers fall into families if they try to emulate the command line
|
|
|
|
interface of another compiler. For example, clang is in the GCC family
|
|
|
|
since it accepts most of the same arguments as GCC. ICL (ICC on
|
|
|
|
windows) is in the MSVC family since it accepts most of the same
|
|
|
|
arguments as MSVC.
|
|
|
|
"""
|
|
|
|
return 'other'
|
|
|
|
|
|
|
|
def get_profile_generate_args(self):
|
|
|
|
raise EnvironmentException(
|
|
|
|
'%s does not support get_profile_generate_args ' % self.get_id())
|
|
|
|
|
|
|
|
def get_profile_use_args(self):
|
|
|
|
raise EnvironmentException(
|
|
|
|
'%s does not support get_profile_use_args ' % self.get_id())
|
|
|
|
|
|
|
|
def get_undefined_link_args(self) -> T.List[str]:
|
|
|
|
return self.linker.get_undefined_link_args()
|
|
|
|
|
|
|
|
def remove_linkerlike_args(self, args):
|
|
|
|
rm_exact = ('-headerpad_max_install_names',)
|
|
|
|
rm_prefixes = ('-Wl,', '-L',)
|
|
|
|
rm_next = ('-L',)
|
|
|
|
ret = []
|
|
|
|
iargs = iter(args)
|
|
|
|
for arg in iargs:
|
|
|
|
# Remove this argument
|
|
|
|
if arg in rm_exact:
|
|
|
|
continue
|
|
|
|
# If the argument starts with this, but is not *exactly* this
|
|
|
|
# f.ex., '-L' should match ['-Lfoo'] but not ['-L', 'foo']
|
|
|
|
if arg.startswith(rm_prefixes) and arg not in rm_prefixes:
|
|
|
|
continue
|
|
|
|
# Ignore this argument and the one after it
|
|
|
|
if arg in rm_next:
|
|
|
|
next(iargs)
|
|
|
|
continue
|
|
|
|
ret.append(arg)
|
|
|
|
return ret
|
|
|
|
|
|
|
|
def get_lto_compile_args(self) -> T.List[str]:
|
|
|
|
return []
|
|
|
|
|
|
|
|
def get_lto_link_args(self) -> T.List[str]:
|
|
|
|
return self.linker.get_lto_args()
|
|
|
|
|
|
|
|
def sanitizer_compile_args(self, value: str) -> T.List[str]:
|
|
|
|
return []
|
|
|
|
|
|
|
|
def sanitizer_link_args(self, value: str) -> T.List[str]:
|
|
|
|
return self.linker.sanitizer_args(value)
|
|
|
|
|
|
|
|
def get_asneeded_args(self) -> T.List[str]:
|
|
|
|
return self.linker.get_asneeded_args()
|
|
|
|
|
|
|
|
def bitcode_args(self) -> T.List[str]:
|
|
|
|
return self.linker.bitcode_args()
|
|
|
|
|
|
|
|
def get_linker_debug_crt_args(self) -> T.List[str]:
|
|
|
|
return self.linker.get_debug_crt_args()
|
|
|
|
|
|
|
|
def get_buildtype_linker_args(self, buildtype: str) -> T.List[str]:
|
|
|
|
return self.linker.get_buildtype_args(buildtype)
|
|
|
|
|
|
|
|
def get_soname_args(self, env: 'Environment', prefix: str, shlib_name: str,
|
|
|
|
suffix: str, soversion: str,
|
|
|
|
darwin_versions: T.Tuple[str, str],
|
|
|
|
is_shared_module: bool) -> T.List[str]:
|
|
|
|
return self.linker.get_soname_args(
|
|
|
|
env, prefix, shlib_name, suffix, soversion,
|
|
|
|
darwin_versions, is_shared_module)
|
|
|
|
|
|
|
|
def get_target_link_args(self, target):
|
|
|
|
return target.link_args
|
|
|
|
|
|
|
|
def get_dependency_compile_args(self, dep):
|
|
|
|
return dep.get_compile_args()
|
|
|
|
|
|
|
|
def get_dependency_link_args(self, dep):
|
|
|
|
return dep.get_link_args()
|
|
|
|
|
|
|
|
@classmethod
|
|
|
|
def use_linker_args(cls, linker: str) -> T.List[str]:
|
|
|
|
"""Get a list of arguments to pass to the compiler to set the linker.
|
|
|
|
"""
|
|
|
|
return []
|
|
|
|
|
|
|
|
|
|
|
|
def get_largefile_args(compiler):
|
|
|
|
'''
|
|
|
|
Enable transparent large-file-support for 32-bit UNIX systems
|
|
|
|
'''
|
|
|
|
if not (compiler.info.is_windows() or compiler.info.is_darwin()):
|
|
|
|
# Enable large-file support unconditionally on all platforms other
|
|
|
|
# than macOS and Windows. macOS is now 64-bit-only so it doesn't
|
|
|
|
# need anything special, and Windows doesn't have automatic LFS.
|
|
|
|
# You must use the 64-bit counterparts explicitly.
|
|
|
|
# glibc, musl, and uclibc, and all BSD libcs support this. On Android,
|
|
|
|
# support for transparent LFS is available depending on the version of
|
|
|
|
# Bionic: https://github.com/android/platform_bionic#32-bit-abi-bugs
|
|
|
|
# https://code.google.com/p/android/issues/detail?id=64613
|
|
|
|
#
|
|
|
|
# If this breaks your code, fix it! It's been 20+ years!
|
|
|
|
return ['-D_FILE_OFFSET_BITS=64']
|
|
|
|
# We don't enable -D_LARGEFILE64_SOURCE since that enables
|
|
|
|
# transitionary features and must be enabled by programs that use
|
|
|
|
# those features explicitly.
|
|
|
|
return []
|
|
|
|
|
|
|
|
|
|
|
|
def get_args_from_envvars(lang: str, use_linker_args: bool) -> T.Tuple[T.List[str], T.List[str]]:
|
|
|
|
"""
|
|
|
|
Returns a tuple of (compile_flags, link_flags) for the specified language
|
|
|
|
from the inherited environment
|
|
|
|
"""
|
|
|
|
def log_var(var, val: T.Optional[str]):
|
|
|
|
if val:
|
|
|
|
mlog.log('Appending {} from environment: {!r}'.format(var, val))
|
|
|
|
else:
|
|
|
|
mlog.debug('No {} in the environment, not changing global flags.'.format(var))
|
|
|
|
|
|
|
|
if lang not in cflags_mapping:
|
|
|
|
return [], []
|
|
|
|
|
|
|
|
compile_flags = [] # type: T.List[str]
|
|
|
|
link_flags = [] # type: T.List[str]
|
|
|
|
|
|
|
|
env_compile_flags = os.environ.get(cflags_mapping[lang])
|
|
|
|
log_var(cflags_mapping[lang], env_compile_flags)
|
|
|
|
if env_compile_flags is not None:
|
|
|
|
compile_flags += split_args(env_compile_flags)
|
|
|
|
|
|
|
|
# Link flags (same for all languages)
|
|
|
|
if lang in languages_using_ldflags:
|
|
|
|
# This is duplicated between the linkers, but I'm not sure how else
|
|
|
|
# to handle this
|
|
|
|
env_link_flags = split_args(os.environ.get('LDFLAGS', ''))
|
|
|
|
else:
|
|
|
|
env_link_flags = []
|
|
|
|
log_var('LDFLAGS', env_link_flags)
|
|
|
|
link_flags += env_link_flags
|
|
|
|
if use_linker_args:
|
|
|
|
# When the compiler is used as a wrapper around the linker (such as
|
|
|
|
# with GCC and Clang), the compile flags can be needed while linking
|
|
|
|
# too. This is also what Autotools does. However, we don't want to do
|
|
|
|
# this when the linker is stand-alone such as with MSVC C/C++, etc.
|
|
|
|
link_flags = compile_flags + link_flags
|
|
|
|
|
|
|
|
# Pre-processor flags for certain languages
|
|
|
|
if lang in {'c', 'cpp', 'objc', 'objcpp'}:
|
|
|
|
env_preproc_flags = os.environ.get('CPPFLAGS')
|
|
|
|
log_var('CPPFLAGS', env_preproc_flags)
|
|
|
|
if env_preproc_flags is not None:
|
|
|
|
compile_flags += split_args(env_preproc_flags)
|
|
|
|
|
|
|
|
return compile_flags, link_flags
|
|
|
|
|
|
|
|
|
|
|
|
def get_global_options(lang: str, comp: T.Type[Compiler],
|
|
|
|
properties: Properties) -> T.Dict[str, coredata.UserOption]:
|
|
|
|
"""Retreive options that apply to all compilers for a given language."""
|
|
|
|
description = 'Extra arguments passed to the {}'.format(lang)
|
|
|
|
opts = {
|
|
|
|
lang + '_args': coredata.UserArrayOption(
|
|
|
|
description + ' compiler',
|
|
|
|
[], split_args=True, user_input=True, allow_dups=True),
|
|
|
|
lang + '_link_args': coredata.UserArrayOption(
|
|
|
|
description + ' linker',
|
|
|
|
[], split_args=True, user_input=True, allow_dups=True),
|
|
|
|
}
|
|
|
|
|
|
|
|
if properties.fallback:
|
|
|
|
# Get from env vars.
|
|
|
|
# XXX: True here is a hack
|
|
|
|
compile_args, link_args = get_args_from_envvars(lang, comp.INVOKES_LINKER)
|
|
|
|
else:
|
|
|
|
compile_args = []
|
|
|
|
link_args = []
|
|
|
|
|
|
|
|
for k, o in opts.items():
|
|
|
|
if k in properties:
|
|
|
|
# Get from configuration files.
|
|
|
|
o.set_value(properties[k])
|
|
|
|
elif k == lang + '_args':
|
|
|
|
o.set_value(compile_args)
|
|
|
|
elif k == lang + '_link_args':
|
|
|
|
o.set_value(link_args)
|
|
|
|
|
|
|
|
return opts
|