glibc/conform/conformtest.py

756 lines
29 KiB
Python
Raw Permalink Normal View History

Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
#!/usr/bin/python3
# Check header contents against the given standard.
# Copyright (C) 2018-2024 Free Software Foundation, Inc.
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
# This file is part of the GNU C Library.
#
# The GNU C Library is free software; you can redistribute it and/or
# modify it under the terms of the GNU Lesser General Public
# License as published by the Free Software Foundation; either
# version 2.1 of the License, or (at your option) any later version.
#
# The GNU C Library is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
# Lesser General Public License for more details.
#
# You should have received a copy of the GNU Lesser General Public
# License along with the GNU C Library; if not, see
Prefer https to http for gnu.org and fsf.org URLs Also, change sources.redhat.com to sourceware.org. This patch was automatically generated by running the following shell script, which uses GNU sed, and which avoids modifying files imported from upstream: sed -ri ' s,(http|ftp)(://(.*\.)?(gnu|fsf|sourceware)\.org($|[^.]|\.[^a-z])),https\2,g s,(http|ftp)(://(.*\.)?)sources\.redhat\.com($|[^.]|\.[^a-z]),https\2sourceware.org\4,g ' \ $(find $(git ls-files) -prune -type f \ ! -name '*.po' \ ! -name 'ChangeLog*' \ ! -path COPYING ! -path COPYING.LIB \ ! -path manual/fdl-1.3.texi ! -path manual/lgpl-2.1.texi \ ! -path manual/texinfo.tex ! -path scripts/config.guess \ ! -path scripts/config.sub ! -path scripts/install-sh \ ! -path scripts/mkinstalldirs ! -path scripts/move-if-change \ ! -path INSTALL ! -path locale/programs/charmap-kw.h \ ! -path po/libc.pot ! -path sysdeps/gnu/errlist.c \ ! '(' -name configure \ -execdir test -f configure.ac -o -f configure.in ';' ')' \ ! '(' -name preconfigure \ -execdir test -f preconfigure.ac ';' ')' \ -print) and then by running 'make dist-prepare' to regenerate files built from the altered files, and then executing the following to cleanup: chmod a+x sysdeps/unix/sysv/linux/riscv/configure # Omit irrelevant whitespace and comment-only changes, # perhaps from a slightly-different Autoconf version. git checkout -f \ sysdeps/csky/configure \ sysdeps/hppa/configure \ sysdeps/riscv/configure \ sysdeps/unix/sysv/linux/csky/configure # Omit changes that caused a pre-commit check to fail like this: # remote: *** error: sysdeps/powerpc/powerpc64/ppc-mcount.S: trailing lines git checkout -f \ sysdeps/powerpc/powerpc64/ppc-mcount.S \ sysdeps/unix/sysv/linux/s390/s390-64/syscall.S # Omit change that caused a pre-commit check to fail like this: # remote: *** error: sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S: last line does not end in newline git checkout -f sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S
2019-09-07 13:40:42 +08:00
# <https://www.gnu.org/licenses/>.
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
import argparse
import fnmatch
import os.path
import re
import subprocess
import sys
import tempfile
import glibcconform
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
class CompileSubTest(object):
"""A compilation subtest."""
def __init__(self, name, text):
"""Initialize a CompileSubTest object."""
self.run_early = False
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.name = name
self.text = text
def run(self, header_tests):
"""Run a compilation subtest."""
header_tests.compile_test(self.name, self.text)
class ExecuteSubTest(object):
"""An execution subtest."""
def __init__(self, name, text):
"""Initialize an ExecuteSubTest object."""
self.run_early = False
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.name = name
self.text = text
def run(self, header_tests):
"""Run an execution subtest."""
header_tests.execute_test(self.name, self.text)
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
class ElementTest(object):
"""Test for an element of a structure or union type."""
def __init__(self, dummy, type_name, member_type, member_name, *rest):
"""Initialize an ElementTest object."""
self.type_name = type_name
self.member_type = member_type
self.member_name = member_name
self.rest = ' '.join(rest)
self.allow_name = self.member_name
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
def gen_subtests(self):
"""Generate subtests for an ElementTest."""
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('%(type_name)s a_%(num)d;\n'
'%(type_name)s b_%(num)d;\n'
'extern void xyzzy_%(num)d '
'(__typeof__ (&b_%(num)d.%(member_name)s), '
'__typeof__ (&a_%(num)d.%(member_name)s), unsigned);\n'
'void foobarbaz_%(num)d (void) {\n'
'xyzzy_%(num)d (&a_%(num)d.%(member_name)s, '
'&b_%(num)d.%(member_name)s, '
'sizeof (a_%(num)d.%(member_name)s));\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'}\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Availability of member %s' % self.member_name,
text))
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('%(type_name)s a2_%(num)d;\n'
'extern %(member_type)s b2_%(num)d%(rest)s;\n'
'extern __typeof__ (a2_%(num)d.%(member_name)s) b2_%(num)d;\n'
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Type of member %s' % self.member_name,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
class ConstantTest(object):
"""Test for a macro or constant."""
def __init__(self, symbol_type, symbol, extra1=None, extra2=None,
extra3=None):
"""Initialize a ConstantTest object."""
self.symbol_type = symbol_type
self.symbol = symbol
# A comparison operation may be specified without a type.
if extra2 is not None and extra3 is None:
self.c_type = None
self.op = extra1
self.value = extra2
else:
self.c_type = extra1
self.op = extra2
self.value = extra3
self.allow_name = self.symbol
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
def gen_subtests(self):
"""Generate subtests for a ConstantTest."""
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
if 'macro' in self.symbol_type:
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('#ifndef %(symbol)s\n'
'# error "Macro %(symbol)s not defined"\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'#endif\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Availability of macro %s' % self.symbol,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
if 'constant' in self.symbol_type:
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('__typeof__ (%(symbol)s) a_%(num)d = %(symbol)s;\n'
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Availability of constant %s' % self.symbol,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
if self.symbol_type == 'macro-int-constant':
sym_bits_def_neg = ''.join(
'# if %s & (1LL << %d)\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
'# define conformtest_%d_bit_%d 0LL\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'# else\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
'# define conformtest_%d_bit_%d (1LL << %d)\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'# endif\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
% (self.symbol, i, self.num, i, self.num, i, i)
for i in range(63))
sym_bits_or_neg = '|'.join('conformtest_%d_bit_%d' % (self.num, i)
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
for i in range(63))
sym_bits_def_pos = ''.join(
'# if %s & (1ULL << %d)\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
'# define conformtest_%d_bit_%d (1ULL << %d)\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'# else\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
'# define conformtest_%d_bit_%d 0ULL\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'# endif\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
% (self.symbol, i, self.num, i, i, self.num, i)
for i in range(64))
sym_bits_or_pos = '|'.join('conformtest_%d_bit_%d' % (self.num, i)
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
for i in range(64))
text = ('#if %s < 0\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
'# define conformtest_%d_negative 1\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'%s'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
'# define conformtest_%d_value ~(%s)\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'#else\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
'# define conformtest_%d_negative 0\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'%s'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
'# define conformtest_%d_value (%s)\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'#endif\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
'_Static_assert (((%s < 0) == conformtest_%d_negative) '
'&& (%s == conformtest_%d_value), '
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'"value match inside and outside #if");\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
% (self.symbol, self.num, sym_bits_def_neg, self.num,
sym_bits_or_neg, self.num, sym_bits_def_pos, self.num,
sym_bits_or_pos, self.symbol, self.num, self.symbol,
self.num))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'#if usability of symbol %s'% self.symbol,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
if self.c_type is not None:
if self.c_type.startswith('promoted:'):
c_type = self.c_type[len('promoted:'):]
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('__typeof__ ((%s) 0 + (%s) 0) a2_%d;\n'
% (c_type, c_type, self.num))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
else:
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = '__typeof__ ((%s) 0) a2_%d;\n' % (self.c_type, self.num)
text += 'extern __typeof__ (%s) a2_%d;\n' % (self.symbol, self.num)
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Type of symbol %s' % self.symbol,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
if self.op is not None:
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('_Static_assert (%(symbol)s %(op)s %(value)s, '
'"value constraint");\n'
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Value of symbol %s' % self.symbol,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
class SymbolTest(object):
"""Test for a symbol (not a compile-time constant)."""
def __init__(self, dummy, symbol, value=None):
"""Initialize a SymbolTest object."""
self.symbol = symbol
self.value = value
self.allow_name = self.symbol
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
def gen_subtests(self):
"""Generate subtests for a SymbolTest."""
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('void foobarbaz_%(num)d (void) {\n'
'__typeof__ (%(symbol)s) a_%(num)d = %(symbol)s;\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'}\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Availability of symbol %s' % self.symbol,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
if self.value is not None:
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('int main (void) { return %(symbol)s != %(symbol)s; }\n'
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(ExecuteSubTest(
'Value of symbol %s' % self.symbol,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
class TypeTest(object):
"""Test for a type name."""
def __init__(self, dummy, type_name):
"""Initialize a TypeTest object."""
self.type_name = type_name
if type_name.startswith('struct '):
self.allow_name = type_name[len('struct '):]
self.maybe_opaque = False
elif type_name.startswith('union '):
self.allow_name = type_name[len('union '):]
self.maybe_opaque = False
else:
self.allow_name = type_name
self.maybe_opaque = True
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
def gen_subtests(self):
"""Generate subtests for a TypeTest."""
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('%s %sa_%d;\n'
% (self.type_name, '*' if self.maybe_opaque else '', self.num))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Availability of type %s' % self.type_name,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
class TagTest(object):
"""Test for a tag name."""
def __init__(self, dummy, type_name):
"""Initialize a TagTest object."""
self.type_name = type_name
if type_name.startswith('struct '):
self.allow_name = type_name[len('struct '):]
elif type_name.startswith('union '):
self.allow_name = type_name[len('union '):]
else:
raise ValueError('unexpected kind of tag: %s' % type_name)
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
def gen_subtests(self):
"""Generate subtests for a TagTest."""
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
# If the tag is not declared, these function prototypes have
# incompatible types.
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('void foo_%(num)d (%(type_name)s *);\n'
'void foo_%(num)d (%(type_name)s *);\n'
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Availability of tag %s' % self.type_name,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
class FunctionTest(object):
"""Test for a function."""
def __init__(self, dummy, return_type, function_name, *args):
"""Initialize a FunctionTest object."""
self.function_name_full = function_name
self.args = ' '.join(args)
if function_name.startswith('(*'):
# Function returning a pointer to function.
self.return_type = '%s (*' % return_type
self.function_name = function_name[len('(*'):]
else:
self.return_type = return_type
self.function_name = function_name
self.allow_name = self.function_name
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
def gen_subtests(self):
"""Generate subtests for a FunctionTest."""
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('%(return_type)s (*foobarbaz_%(num)d) %(args)s '
'= %(function_name)s;\n'
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Availability of function %s' % self.function_name,
text))
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('extern %(return_type)s (*foobarbaz2_%(num)d) %(args)s;\n'
'extern __typeof__ (&%(function_name)s) foobarbaz2_%(num)d;\n'
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Type of function %s' % self.function_name,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
class VariableTest(object):
"""Test for a variable."""
def __init__(self, dummy, var_type, var_name, *rest):
"""Initialize a VariableTest object."""
self.var_type = var_type
self.var_name = var_name
self.rest = ' '.join(rest)
self.allow_name = var_name
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
def gen_subtests(self):
"""Generate subtests for a VariableTest."""
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('typedef %(var_type)s xyzzy_%(num)d%(rest)s;\n'
'xyzzy_%(num)d *foobarbaz_%(num)d = &%(var_name)s;\n'
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Availability of variable %s' % self.var_name,
text))
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('extern %(var_type)s %(var_name)s%(rest)s;\n'
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Type of variable %s' % self.var_name,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
class MacroFunctionTest(object):
"""Test for a possibly macro-only function."""
def __init__(self, dummy, return_type, function_name, *args):
"""Initialize a MacroFunctionTest object."""
self.return_type = return_type
self.function_name = function_name
self.args = ' '.join(args)
self.allow_name = function_name
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
def gen_subtests(self):
"""Generate subtests for a MacroFunctionTest."""
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('#ifndef %(function_name)s\n'
'%(return_type)s (*foobarbaz_%(num)d) %(args)s '
'= %(function_name)s;\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'#endif\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Availability of macro %s' % self.function_name,
text))
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('#ifndef %(function_name)s\n'
'extern %(return_type)s (*foobarbaz2_%(num)d) %(args)s;\n'
'extern __typeof__ (&%(function_name)s) foobarbaz2_%(num)d;\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'#endif\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Type of macro %s' % self.function_name,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
class MacroStrTest(object):
"""Test for a string-valued macro."""
def __init__(self, dummy, macro_name, value):
"""Initialize a MacroStrTest object."""
self.macro_name = macro_name
self.value = value
self.allow_name = macro_name
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
def gen_subtests(self):
"""Generate subtests for a MacroStrTest."""
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
text = ('#ifndef %(macro_name)s\n'
'# error "Macro %(macro_name)s not defined"\n'
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
'#endif\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(CompileSubTest(
'Availability of macro %s' % self.macro_name,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
# We can't include <string.h> here.
text = ('extern int (strcmp)(const char *, const char *);\n'
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
'int main (void) { return (strcmp) (%(macro_name)s, '
'%(value)s) != 0; }\n'
% vars(self))
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
self.subtests.append(ExecuteSubTest(
'Value of macro %s' % self.macro_name,
text))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
class HeaderTests(object):
"""The set of tests run for a header."""
Run conform/ tests using newly built libc Although the conform/ header tests are built using the headers of the glibc under test, the execution tests from conformtest (a few tests of the values of macros evaluating to string constants) are linked and run with system libc, not the newly built libc. Apart from preventing testing in cross environments, this can be a problem even for native testing. Specifically, it can be useful to do native testing when building with a cross compiler that links with a libc that is not the system libc; for example, on x86_64, you can test all three ABIs that way if the kernel support is present, even if the host OS lacks 32-bit or x32 libraries or they are older than the libraries in the sysroot used by the compiler used to build glibc. This works for almost all tests, but not for these conformtest tests. Arrange for conformtest to link and run test programs similarly to other tests, with consequent refactoring of various variables in Makeconfig to allow passing relevant parts of the link-time command lines down to conformtest. In general, the parts of the link command involving $@ or $^ are separated out from the parts that should be passed to conformtest (the variables passed to conformtest still involve various variables whose names involve $(@F), but those variables simply won't be defined for the conformtest makefile rules and I think their presence there is harmless). This is also most of the support that would be needed to allow running those tests of string constants for cross testing when test-wrapper is defined. That will also need changes to where conformtest.py puts the test executables, so it puts them in the main object directory (expected to be shared with a test system in cross testing) rather than /tmp (not expected to be shared) as at present. Tested for x86_64.
2021-09-17 21:12:10 +08:00
def __init__(self, header, standard, cc, flags, ldflags, libs,
run_program_prefix, cross, xfail):
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
"""Initialize a HeaderTests object."""
self.header = header
self.standard = standard
self.cc = cc
self.flags = flags
Run conform/ tests using newly built libc Although the conform/ header tests are built using the headers of the glibc under test, the execution tests from conformtest (a few tests of the values of macros evaluating to string constants) are linked and run with system libc, not the newly built libc. Apart from preventing testing in cross environments, this can be a problem even for native testing. Specifically, it can be useful to do native testing when building with a cross compiler that links with a libc that is not the system libc; for example, on x86_64, you can test all three ABIs that way if the kernel support is present, even if the host OS lacks 32-bit or x32 libraries or they are older than the libraries in the sysroot used by the compiler used to build glibc. This works for almost all tests, but not for these conformtest tests. Arrange for conformtest to link and run test programs similarly to other tests, with consequent refactoring of various variables in Makeconfig to allow passing relevant parts of the link-time command lines down to conformtest. In general, the parts of the link command involving $@ or $^ are separated out from the parts that should be passed to conformtest (the variables passed to conformtest still involve various variables whose names involve $(@F), but those variables simply won't be defined for the conformtest makefile rules and I think their presence there is harmless). This is also most of the support that would be needed to allow running those tests of string constants for cross testing when test-wrapper is defined. That will also need changes to where conformtest.py puts the test executables, so it puts them in the main object directory (expected to be shared with a test system in cross testing) rather than /tmp (not expected to be shared) as at present. Tested for x86_64.
2021-09-17 21:12:10 +08:00
self.ldflags = ldflags
self.libs = libs
self.run_program_prefix = run_program_prefix
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
self.cross = cross
self.xfail_str = xfail
self.cflags_namespace = ('%s -fno-builtin %s -D_ISOMAC'
% (flags, glibcconform.CFLAGS[standard]))
# When compiling the conformance test programs, use of
# __attribute__ in headers is disabled because of attributes
# that affect the types of functions as seen by typeof.
self.cflags = "%s '-D__attribute__(x)='" % self.cflags_namespace
self.tests = []
self.allow = set()
self.allow_fnmatch = set()
self.headers_handled = set()
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
self.num_tests = 0
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
self.total = 0
self.skipped = 0
self.errors = 0
self.xerrors = 0
def add_allow(self, name, pattern_ok):
"""Add an identifier as an allowed token for this header.
If pattern_ok, fnmatch patterns are OK as well as
identifiers.
"""
if re.fullmatch(r'[A-Za-z_][A-Za-z0-9_]*', name):
self.allow.add(name)
elif pattern_ok:
self.allow_fnmatch.add(name)
else:
raise ValueError('bad identifier: %s' % name)
def check_token(self, bad_tokens, token):
"""Check whether an identifier token is allowed, and record it in
bad_tokens if not.
"""
if token.startswith('_'):
return
if token in glibcconform.KEYWORDS[self.standard]:
return
if token in self.allow:
return
for pattern in self.allow_fnmatch:
if fnmatch.fnmatch(token, pattern):
return
bad_tokens.add(token)
def handle_test_line(self, line, allow):
"""Handle a single line in the test data.
If allow is true, the header is one specified in allow-header
and so tests are marked as allowed for namespace purposes but
otherwise ignored.
"""
orig_line = line
xfail = False
if line.startswith('xfail-'):
xfail = True
line = line[len('xfail-'):]
else:
match = re.match(r'xfail\[(.*?)\]-(.*)', line)
if match:
xfail_cond = match.group(1)
line = match.group(2)
# "xfail[cond]-" or "xfail[cond1|cond2|...]-" means a
# failure of the test is allowed if any of the listed
# conditions are in the --xfail command-line option
# argument.
if self.xfail_str and re.search(r'\b(%s)\b' % xfail_cond,
self.xfail_str):
xfail = True
optional = False
if line.startswith('optional-'):
optional = True
line = line[len('optional-'):]
# Tokens in test data are space-separated, except for {...}
# tokens that may contain spaces.
tokens = []
while line:
match = re.match(r'\{(.*?)\}(.*)', line)
if match:
tokens.append(match.group(1))
line = match.group(2)
else:
match = re.match(r'([^ ]*)(.*)', line)
tokens.append(match.group(1))
line = match.group(2)
line = line.strip()
if tokens[0] == 'allow-header':
if len(tokens) != 2 or xfail or optional:
raise ValueError('bad allow-header line: %s' % orig_line)
if tokens[1] not in self.headers_handled:
self.load_tests(tokens[1], True)
return
if tokens[0] == 'allow':
if len(tokens) != 2 or xfail or optional:
raise ValueError('bad allow line: %s' % orig_line)
self.add_allow(tokens[1], True)
return
test_classes = {'element': ElementTest,
'macro': ConstantTest,
'constant': ConstantTest,
'macro-constant': ConstantTest,
'macro-int-constant': ConstantTest,
'symbol': SymbolTest,
'type': TypeTest,
'tag': TagTest,
'function': FunctionTest,
'variable': VariableTest,
'macro-function': MacroFunctionTest,
'macro-str': MacroStrTest}
test = test_classes[tokens[0]](*tokens)
test.xfail = xfail
test.optional = optional
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
test.num = self.num_tests
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
test.subtests = []
Use unique identifiers in conformtest. This patch makes tests in conformtest use unique identifiers, in preparation for trying to cover more tests in a single compilation to speed up these tests as suggested in <https://sourceware.org/ml/libc-alpha/2018-11/msg00229.html>. Tests are assigned a number, used in generating identifiers; where a single call to a run method does multiple compilations (sharing that number), identifiers are changed as needed to avoid duplication between those compilations, so they can be combined in future. Large numbers of positional arguments to format strings make the code harder to follow, and using the test numbers serves to increase the number of arguments to such format strings, so the code is generally changed to use %(name)s where all the arguments come from attributes of the test object and so vars(self) is sufficient to provide all those names for the format string. Cases where some arguments aren't attributes of self still use positional format arguments. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (ElementTest.run): Use unique identifiers in tests. Use names for format arguments. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.__init__): Set self.num_tests. (HeaderTests.handle_test_line): Set test.num. Increment self.num_tests.
2018-11-20 11:54:53 +08:00
self.num_tests += 1
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
self.add_allow(test.allow_name, False)
if not allow:
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
test.gen_subtests()
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
self.tests.append(test)
def load_tests(self, header, allow):
"""Load tests of a header.
If allow is true, the header is one specified in allow-header
and so tests are marked as allowed for namespace purposes but
otherwise ignored.
"""
self.headers_handled.add(header)
header_s = header.replace('/', '_')
temp_file = os.path.join(self.temp_dir, 'header-data-%s' % header_s)
cmd = ('%s -E -D%s -std=c99 -x c data/%s-data > %s'
% (self.cc, self.standard, header, temp_file))
subprocess.check_call(cmd, shell=True)
with open(temp_file, 'r') as tests:
for line in tests:
line = line.strip()
if line == '' or line.startswith('#'):
continue
self.handle_test_line(line, allow)
def note_error(self, name, xfail):
"""Note a failing test."""
if xfail:
print('XFAIL: %s' % name)
self.xerrors += 1
else:
print('FAIL: %s' % name)
self.errors += 1
sys.stdout.flush()
def note_skip(self, name):
"""Note a skipped test."""
print('SKIP: %s' % name)
self.skipped += 1
sys.stdout.flush()
def compile_test(self, name, text):
"""Run a compilation test; return True if it passes."""
self.total += 1
if self.group_ignore:
return False
optional = self.group_optional
self.group_optional = False
if self.group_skip:
self.note_skip(name)
return False
c_file = os.path.join(self.temp_dir, 'test.c')
o_file = os.path.join(self.temp_dir, 'test.o')
with open(c_file, 'w') as c_file_out:
c_file_out.write('#include <%s>\n%s' % (self.header, text))
cmd = ('%s %s -c %s -o %s' % (self.cc, self.cflags, c_file, o_file))
try:
subprocess.check_call(cmd, shell=True)
except subprocess.CalledProcessError:
if optional:
print('MISSING: %s' % name)
sys.stdout.flush()
self.group_ignore = True
else:
self.note_error(name, self.group_xfail)
self.group_skip = True
return False
print('PASS: %s' % name)
sys.stdout.flush()
return True
def execute_test(self, name, text):
"""Run an execution test."""
self.total += 1
if self.group_ignore:
return False
if self.group_skip:
self.note_skip(name)
return
c_file = os.path.join(self.temp_dir, 'test.c')
exe_file = os.path.join(self.temp_dir, 'test')
with open(c_file, 'w') as c_file_out:
c_file_out.write('#include <%s>\n%s' % (self.header, text))
Run conform/ tests using newly built libc Although the conform/ header tests are built using the headers of the glibc under test, the execution tests from conformtest (a few tests of the values of macros evaluating to string constants) are linked and run with system libc, not the newly built libc. Apart from preventing testing in cross environments, this can be a problem even for native testing. Specifically, it can be useful to do native testing when building with a cross compiler that links with a libc that is not the system libc; for example, on x86_64, you can test all three ABIs that way if the kernel support is present, even if the host OS lacks 32-bit or x32 libraries or they are older than the libraries in the sysroot used by the compiler used to build glibc. This works for almost all tests, but not for these conformtest tests. Arrange for conformtest to link and run test programs similarly to other tests, with consequent refactoring of various variables in Makeconfig to allow passing relevant parts of the link-time command lines down to conformtest. In general, the parts of the link command involving $@ or $^ are separated out from the parts that should be passed to conformtest (the variables passed to conformtest still involve various variables whose names involve $(@F), but those variables simply won't be defined for the conformtest makefile rules and I think their presence there is harmless). This is also most of the support that would be needed to allow running those tests of string constants for cross testing when test-wrapper is defined. That will also need changes to where conformtest.py puts the test executables, so it puts them in the main object directory (expected to be shared with a test system in cross testing) rather than /tmp (not expected to be shared) as at present. Tested for x86_64.
2021-09-17 21:12:10 +08:00
cmd = ('%s %s %s %s %s -o %s' % (self.cc, self.cflags, self.ldflags,
c_file, self.libs, exe_file))
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
try:
subprocess.check_call(cmd, shell=True)
except subprocess.CalledProcessError:
self.note_error(name, self.group_xfail)
return
if self.cross:
self.note_skip(name)
return
try:
Run conform/ tests using newly built libc Although the conform/ header tests are built using the headers of the glibc under test, the execution tests from conformtest (a few tests of the values of macros evaluating to string constants) are linked and run with system libc, not the newly built libc. Apart from preventing testing in cross environments, this can be a problem even for native testing. Specifically, it can be useful to do native testing when building with a cross compiler that links with a libc that is not the system libc; for example, on x86_64, you can test all three ABIs that way if the kernel support is present, even if the host OS lacks 32-bit or x32 libraries or they are older than the libraries in the sysroot used by the compiler used to build glibc. This works for almost all tests, but not for these conformtest tests. Arrange for conformtest to link and run test programs similarly to other tests, with consequent refactoring of various variables in Makeconfig to allow passing relevant parts of the link-time command lines down to conformtest. In general, the parts of the link command involving $@ or $^ are separated out from the parts that should be passed to conformtest (the variables passed to conformtest still involve various variables whose names involve $(@F), but those variables simply won't be defined for the conformtest makefile rules and I think their presence there is harmless). This is also most of the support that would be needed to allow running those tests of string constants for cross testing when test-wrapper is defined. That will also need changes to where conformtest.py puts the test executables, so it puts them in the main object directory (expected to be shared with a test system in cross testing) rather than /tmp (not expected to be shared) as at present. Tested for x86_64.
2021-09-17 21:12:10 +08:00
subprocess.check_call('%s %s' % (self.run_program_prefix,
exe_file),
shell=True)
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
except subprocess.CalledProcessError:
self.note_error(name, self.group_xfail)
return
print('PASS: %s' % name)
sys.stdout.flush()
def check_namespace(self, name):
"""Check the namespace of a header."""
c_file = os.path.join(self.temp_dir, 'namespace.c')
out_file = os.path.join(self.temp_dir, 'namespace-out')
with open(c_file, 'w') as c_file_out:
c_file_out.write('#include <%s>\n' % self.header)
cmd = ('%s %s -E %s -P -Wp,-dN > %s'
% (self.cc, self.cflags_namespace, c_file, out_file))
subprocess.check_call(cmd, shell=True)
bad_tokens = set()
with open(out_file, 'r') as content:
for line in content:
line = line.strip()
if not line:
continue
if re.match(r'# [1-9]', line):
continue
if line.startswith('#pragma GCC '):
# No GCC pragma uses macro expansion, so no
# namespace issues arise from such pragmas. (Some
# pragmas not in the GCC namespace do macro-expand
# their arguments and so could be affected by
# macros defined by user code including the
# header.)
continue
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
match = re.match(r'#define (.*)', line)
if match:
self.check_token(bad_tokens, match.group(1))
continue
match = re.match(r'#undef (.*)', line)
if match:
bad_tokens.discard(match.group(1))
continue
# Tokenize the line and check identifiers found. The
# handling of strings and character constants does not
# allow for escaped quotes, and hex floats may be
# wrongly split into tokens including identifiers, but
# this is sufficient in practice.
line = re.sub(r'(?:\bL)?(?:"[^"]*"|\'[^\']*\')', '', line)
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
line = line.strip()
for token in re.split(r'[^A-Za-z0-9_]+', line):
if re.match(r'[A-Za-z_]', token):
self.check_token(bad_tokens, token)
if bad_tokens:
for token in sorted(bad_tokens):
print(' Namespace violation: "%s"' % token)
self.note_error(name, False)
else:
print('PASS: %s' % name)
sys.stdout.flush()
def run(self):
"""Load and run tests of a header."""
with tempfile.TemporaryDirectory() as self.temp_dir:
self.load_tests(self.header, False)
self.group_optional = False
self.group_xfail = False
self.group_ignore = False
self.group_skip = False
available = self.compile_test('Availability of <%s>' % self.header,
'')
if available:
# As an optimization, try running all non-optional,
# non-XFAILed compilation tests in a single execution
# of the compiler.
combined_list = []
for test in self.tests:
if not test.optional and not test.xfail:
for subtest in test.subtests:
if isinstance(subtest, CompileSubTest):
combined_list.append(subtest.text)
subtest.run_early = True
combined_ok = self.compile_test('Combined <%s> test'
% self.header,
'\n'.join(combined_list))
# Now run the other tests, or all tests if the
# combined test failed.
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
for test in self.tests:
# A test may run more than one subtest. If the
# initial subtest for an optional symbol fails,
# others are not run at all; if the initial
# subtest for an optional symbol succeeds, others
# are run and are not considered optional; if the
# initial subtest for a required symbol fails,
# others are skipped.
self.group_optional = test.optional
self.group_xfail = test.xfail
self.group_ignore = False
self.group_skip = False
Separate conformtest subtest generation and execution. This patch continues moving conformtest towards running more tests in a single compiler execution by separating the generation and execution of the subtests of each test. Instead of test classes having a run method that both generates the text of the programs to be compiled or executed, and compiles or executes them, they are changed to having a gen_subtests method that just generates CompileSubTest and ExecuteSubTest objects to store the subtest names and text, and then a separate loop in HeaderTests.run deals with actually executing those subtests. This will allow for future changes to extract the text for all non-optional, non-xfail compilation subtests to try compiling those all at once, with separate compilations only if that fails, so massively reducing the number of separate compiler executions (each of which needs to parse the entire contents of the header under test, in addition to the startup cost that applies even for compiling an empty file). Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py (CompileSubTest): New class. (ExecuteSubTest): Likewise. (ElementTest.run): Rename to gen_subtests. Append tests to self.subtests instead of running them. (ConstantTest.run): Likewise. (SymbolTest.run): Likewise. (TypeTest.run): Likewise. (TagTest.run): Likewise. (FunctionTest.run): Likewise. (VariableTest.run): Likewise. (MacroFunctionTest.run): Likewise. (MacroStrTest.run): Likewise. (HeaderTests.handle_test_line): Generate subtests for tests. (HeaderTests.run): Run subtests for tests.
2018-11-22 08:47:20 +08:00
for subtest in test.subtests:
if combined_ok and subtest.run_early:
self.total += 1
print('PASSCOMBINED: %s' % subtest.name)
sys.stdout.flush()
else:
subtest.run(self)
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
namespace_name = 'Namespace of <%s>' % self.header
if available:
self.check_namespace(namespace_name)
else:
self.note_skip(namespace_name)
print('-' * 76)
print(' Total number of tests : %4d' % self.total)
print(' Number of failed tests : %4d' % self.errors)
print(' Number of xfailed tests : %4d' % self.xerrors)
print(' Number of skipped tests : %4d' % self.skipped)
sys.exit(1 if self.errors else 0)
def main():
"""The main entry point."""
parser = argparse.ArgumentParser(description='Check header contents.')
parser.add_argument('--header', metavar='HEADER',
help='name of header')
parser.add_argument('--standard', metavar='STD',
help='standard to use when processing header')
parser.add_argument('--cc', metavar='CC',
help='C compiler to use')
parser.add_argument('--flags', metavar='CFLAGS',
help='Compiler flags to use with CC')
Run conform/ tests using newly built libc Although the conform/ header tests are built using the headers of the glibc under test, the execution tests from conformtest (a few tests of the values of macros evaluating to string constants) are linked and run with system libc, not the newly built libc. Apart from preventing testing in cross environments, this can be a problem even for native testing. Specifically, it can be useful to do native testing when building with a cross compiler that links with a libc that is not the system libc; for example, on x86_64, you can test all three ABIs that way if the kernel support is present, even if the host OS lacks 32-bit or x32 libraries or they are older than the libraries in the sysroot used by the compiler used to build glibc. This works for almost all tests, but not for these conformtest tests. Arrange for conformtest to link and run test programs similarly to other tests, with consequent refactoring of various variables in Makeconfig to allow passing relevant parts of the link-time command lines down to conformtest. In general, the parts of the link command involving $@ or $^ are separated out from the parts that should be passed to conformtest (the variables passed to conformtest still involve various variables whose names involve $(@F), but those variables simply won't be defined for the conformtest makefile rules and I think their presence there is harmless). This is also most of the support that would be needed to allow running those tests of string constants for cross testing when test-wrapper is defined. That will also need changes to where conformtest.py puts the test executables, so it puts them in the main object directory (expected to be shared with a test system in cross testing) rather than /tmp (not expected to be shared) as at present. Tested for x86_64.
2021-09-17 21:12:10 +08:00
parser.add_argument('--ldflags', metavar='LDFLAGS',
help='Compiler arguments for linking before inputs')
parser.add_argument('--libs', metavar='LIBS',
help='Compiler arguments for linking after inputs')
parser.add_argument('--run-program-prefix', metavar='RUN-PROGRAM-PREFIX',
help='Wrapper for running newly built program')
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
parser.add_argument('--cross', action='store_true',
help='Do not run compiled test programs')
parser.add_argument('--xfail', metavar='COND',
help='Name of condition for XFAILs')
args = parser.parse_args()
tests = HeaderTests(args.header, args.standard, args.cc, args.flags,
Run conform/ tests using newly built libc Although the conform/ header tests are built using the headers of the glibc under test, the execution tests from conformtest (a few tests of the values of macros evaluating to string constants) are linked and run with system libc, not the newly built libc. Apart from preventing testing in cross environments, this can be a problem even for native testing. Specifically, it can be useful to do native testing when building with a cross compiler that links with a libc that is not the system libc; for example, on x86_64, you can test all three ABIs that way if the kernel support is present, even if the host OS lacks 32-bit or x32 libraries or they are older than the libraries in the sysroot used by the compiler used to build glibc. This works for almost all tests, but not for these conformtest tests. Arrange for conformtest to link and run test programs similarly to other tests, with consequent refactoring of various variables in Makeconfig to allow passing relevant parts of the link-time command lines down to conformtest. In general, the parts of the link command involving $@ or $^ are separated out from the parts that should be passed to conformtest (the variables passed to conformtest still involve various variables whose names involve $(@F), but those variables simply won't be defined for the conformtest makefile rules and I think their presence there is harmless). This is also most of the support that would be needed to allow running those tests of string constants for cross testing when test-wrapper is defined. That will also need changes to where conformtest.py puts the test executables, so it puts them in the main object directory (expected to be shared with a test system in cross testing) rather than /tmp (not expected to be shared) as at present. Tested for x86_64.
2021-09-17 21:12:10 +08:00
args.ldflags, args.libs, args.run_program_prefix,
Replace conformtest.pl with conformtest.py. Continuing the consolidation on Python for various miscellaneous build and test scripts, this patch moves conformtest from Perl to Python. The substance of the tests run is intended to be the same as before, except that the previous test for tags did not actually achieve the intended purpose of verifying whether a tag was already declared, so is changed to one that would actually fail for a tag that wasn't declared, and a typo in the old test for variables being available ($xyzzy instead of xyzzy) would have made that test not use the correct type (but it would have passed anyway with warnings). No attempt is made to keep the details of what the test output looks like; instead, tests are given names which are made to follow PASS: / FAIL: / XFAIL: / SKIP: / MISSING: as appropriate. In the new version, there is more consistent parsing of test lines (into a series of words, either surrounded by {} or separated by spaces) that applies for all kinds of test lines, rather than the old approach of different regular expressions for every kind of test. A few of the conform/data/ files are adjusted so their syntax works with the new script (which now requires spaces in certain cases where the old script tolerated them being missing, and does not allow stray semicolons at the end of "function" lines). Similarly, common logic is used around what happens with a second subtest if a first one fails (e.g., a test for a symbol's type if the test for availability fails), rather than such logic being replicated separately for each kind of test. Common parsing also applies for test lines both when they are lines for the header under test and when they are lines for another header specified with allow-header, again unlike the old script. Tested for x86_64, and with build-many-glibcs.py. * conform/conformtest.py: New file. * conform/conformtest.pl: Remove. * conform/GlibcConform.pm: Likewise. * conform/glibcconform.py (KEYWORDS_C90): New constant. (KEYWORDS_C99): Likewise. (KEYWORDS): Likewise. * conform/Makefile ($(conformtest-header-tests)): Use conformtest.py instead of conformtest.pl. Do not pass --tmpdir option. Use --header instead of --headers. * conform/data/arpa/inet.h-data: Remove trailing semicolons on function entries. * conform/data/spawn.h-data: Likewise. * conform/data/fcntl.h-data (openat): Add space after function name. * conform/data/wchar.h-data (wcscasecmp): Likewise. (wcscasecmp_l): Likewise. * conform/data/termios.h-data (c_cc): Add space after element name.
2018-11-09 21:54:06 +08:00
args.cross, args.xfail)
tests.run()
if __name__ == '__main__':
main()