binutils-gdb/gdb/unittests/enum-flags-selftests.c

597 lines
16 KiB
C
Raw Normal View History

Rewrite enum_flags, add unit tests, fix problems This patch started by adding comprehensive unit tests for enum_flags. For the testing part, it adds: - tests of normal expected uses of the API. - checks that _invalid_ uses of the API would fail to compile. I.e., it validates that enum_flags really is a strong type, and that incorrect mixing of enum types would be caught at compile time. It pulls that off making use of SFINEA and C++11's decltype/constexpr. This revealed many holes in the enum_flags API. For example, the f1 assignment below currently incorrectly fails to compile: enum_flags<flags> f1 = FLAG1; enum_flags<flags> f2 = FLAG2 | f1; The unit tests also revealed that this useful use case doesn't work: enum flag { FLAG1 = 1, FLAG2 = 2 }; enum_flags<flag> src = FLAG1; enum_flags<flag> f1 = condition ? src : FLAG2; It fails to compile because enum_flags<flag> and flag are convertible to each other. Turns out that making enum_flags be implicitly convertible to the backing raw enum type was not a good idea. If we make it convertible to the underlying type instead, we fix that ternary operator use case, and, we find cases throughout the codebase that should be using the enum_flags but were using the raw backing enum instead. So it's a good change overall. Also, several operators were missing. These holes and more are plugged by this patch, by reworking how the enum_flags operators are implemented, and making use of C++11's feature of being able to delete methods/functions. There are cases in gdb/compile/ where we need to call a function in a C plugin API that expects the raw enum. To address cases like that, this adds a "raw()" method to enum_flags. This way we can keep using the safer enum_flags to construct the value, and then be explicit when we need to get at the raw enum. This makes most of the enum_flags operators constexpr. Beyond enabling more compiler optimizations and enabling the new unit tests, this has other advantages, like making it possible to use operator| with enum_flags values in switch cases, where only compile-time constants are allowed: enum_flags<flags> f = FLAG1 | FLAG2; switch (f) { case FLAG1 | FLAG2: break; } Currently that fails to compile. It also switches to a different mechanism of enabling the global operators. The current mechanism isn't namespace friendly, the new one is. It also switches to C++11-style SFINAE -- instead of wrapping the return type in a SFINAE-friently structure, we use an unnamed template parameter. I.e., this: template <typename enum_type, typename = is_enum_flags_enum_type_t<enum_type>> enum_type operator& (enum_type e1, enum_type e2) instead of: template <typename enum_type> typename enum_flags_type<enum_type>::type operator& (enum_type e1, enum_type e2) Note that the static_assert inside operator~() was converted to a couple overloads (signed vs unsigned), because static_assert is too late for SFINAE-based tests, which is important for the CHECK_VALID unit tests. Tested with gcc {4.8, 7.1, 9.3} and clang {5.0.2, 10.0.0}. gdb/ChangeLog: * Makefile.in (SELFTESTS_SRCS): Add unittests/enum-flags-selftests.c. * btrace.c (ftrace_update_caller, ftrace_fixup_calle): Use btrace_function_flags instead of enum btrace_function_flag. * compile/compile-c-types.c (convert_qualified): Use enum_flags::raw. * compile/compile-cplus-symbols.c (convert_one_symbol) (convert_symbol_bmsym): * compile/compile-cplus-types.c (compile_cplus_convert_method) (compile_cplus_convert_struct_or_union_methods) (compile_cplus_instance::convert_qualified_base): * go-exp.y (parse_string_or_char): Add cast to int. * unittests/enum-flags-selftests.c: New file. * record-btrace.c (btrace_thread_flag_to_str): Change parameter's type to btrace_thread_flags from btrace_thread_flag. (record_btrace_cancel_resume, record_btrace_step_thread): Change local's type to btrace_thread_flags from btrace_thread_flag. Add cast in DEBUG call. gdbsupport/ChangeLog: * enum-flags.h: Include "traits.h". (DEF_ENUM_FLAGS_TYPE): Declare a function instead of defining a structure. (enum_underlying_type): Update comment. (namespace enum_flags_detail): New. Move struct zero_type here. (EnumIsUnsigned, EnumIsSigned): New. (class enum_flags): Make most methods constexpr. (operator&=, operator|=, operator^=): Take an enum_flags instead of an enum_type. Make rvalue ref versions deleted. (operator enum_type()): Delete. (operator&, operator|, operator^, operator~): Delete, moved out of class. (raw()): New method. (is_enum_flags_enum_type_t): Declare. (ENUM_FLAGS_GEN_BINOP, ENUM_FLAGS_GEN_COMPOUND_ASSIGN) (ENUM_FLAGS_GEN_COMP): New. Use them to reimplement global operators. (operator~): Now constexpr and reimplemented. (operator<<, operator>>): New deleted functions. * valid-expr.h (CHECK_VALID_EXPR_5, CHECK_VALID_EXPR_6): New.
2020-09-15 04:16:59 +08:00
/* Self tests for enum-flags for GDB, the GNU debugger.
Copyright (C) 2016-2020 Free Software Foundation, Inc.
This file is part of GDB.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 3 of the License, or
(at your option) any later version.
This program 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 General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>. */
#include "defs.h"
#include "gdbsupport/enum-flags.h"
#include "gdbsupport/valid-expr.h"
#include "gdbsupport/selftest.h"
namespace selftests {
namespace enum_flags_tests {
/* The (real) enum types used in CHECK_VALID. Their names match the
template parameter names of the templates defined by CHECK_VALID to
make it simpler to use. They could be named differently. */
/* A "real enum". */
enum RE
{
RE_FLAG1 = 1 << 1,
RE_FLAG2 = 1 << 2,
};
/* Another "real enum". */
enum RE2
{
RE2_FLAG1 = 1 << 1,
RE2_FLAG2 = 1 << 2,
};
/* An unsigned "real enum". */
enum URE : unsigned
{
URE_FLAG1 = 1 << 1,
URE_FLAG2 = 1 << 2,
URE_FLAG3 = 0xffffffff,
};
/* A non-flags enum. */
enum NF
{
NF_FLAG1 = 1 << 1,
NF_FLAG2 = 1 << 2,
};
/* The corresponding "enum flags" types. */
DEF_ENUM_FLAGS_TYPE (RE, EF);
DEF_ENUM_FLAGS_TYPE (RE2, EF2);
DEF_ENUM_FLAGS_TYPE (URE, UEF);
#if HAVE_IS_TRIVIALLY_COPYABLE
/* So that std::vectors of types that have enum_flags fields can
reallocate efficiently memcpy. */
gdb_static_assert (std::is_trivially_copyable<EF>::value);
#endif
/* A couple globals used as lvalues in the CHECK_VALID expressions
below. Their names (and types) match the uppercase type names
exposed by CHECK_VALID just to make the expressions easier to
follow. */
static RE re ATTRIBUTE_UNUSED;
static EF ef ATTRIBUTE_UNUSED;
/* First, compile-time tests that:
- make sure that incorrect operations with mismatching enum types
are caught at compile time.
- make sure that the same operations but involving the right enum
types do compile and that they return the correct type.
*/
#define CHECK_VALID(VALID, EXPR_TYPE, EXPR) \
CHECK_VALID_EXPR_6 (EF, RE, EF2, RE2, UEF, URE, VALID, EXPR_TYPE, EXPR)
typedef std::underlying_type<RE>::type und;
/* Test construction / conversion from/to different types. */
/* RE/EF -> underlying (explicit) */
CHECK_VALID (true, und, und (RE ()))
CHECK_VALID (true, und, und (EF ()))
/* RE/EF -> int (explicit) */
CHECK_VALID (true, int, int (RE ()))
CHECK_VALID (true, int, int (EF ()))
/* other -> RE */
/* You can construct a raw enum value from an int explicitly to punch
a hole in the type system if need to. */
CHECK_VALID (true, RE, RE (1))
CHECK_VALID (true, RE, RE (RE2 ()))
CHECK_VALID (false, void, RE (EF2 ()))
CHECK_VALID (true, RE, RE (RE ()))
CHECK_VALID (false, void, RE (EF ()))
/* other -> EF. */
/* As expected, enum-flags is a stronger type than the backing raw
enum. Unlike with raw enums, you can't construct an enum flags
from an integer nor from an unrelated enum type explicitly. Add an
intermediate conversion via the raw enum if you really need it. */
CHECK_VALID (false, void, EF (1))
CHECK_VALID (false, void, EF (1u))
CHECK_VALID (false, void, EF (RE2 ()))
CHECK_VALID (false, void, EF (EF2 ()))
CHECK_VALID (true, EF, EF (RE ()))
CHECK_VALID (true, EF, EF (EF ()))
/* Test operators. */
/* operator OP (raw_enum, int) */
CHECK_VALID (false, void, RE () | 1)
CHECK_VALID (false, void, RE () & 1)
CHECK_VALID (false, void, RE () ^ 1)
/* operator OP (int, raw_enum) */
CHECK_VALID (false, void, 1 | RE ())
CHECK_VALID (false, void, 1 & RE ())
CHECK_VALID (false, void, 1 ^ RE ())
/* operator OP (enum_flags, int) */
CHECK_VALID (false, void, EF () | 1)
CHECK_VALID (false, void, EF () & 1)
CHECK_VALID (false, void, EF () ^ 1)
/* operator OP (int, enum_flags) */
CHECK_VALID (false, void, 1 | EF ())
CHECK_VALID (false, void, 1 & EF ())
CHECK_VALID (false, void, 1 ^ EF ())
/* operator OP (raw_enum, raw_enum) */
CHECK_VALID (false, void, RE () | RE2 ())
CHECK_VALID (false, void, RE () & RE2 ())
CHECK_VALID (false, void, RE () ^ RE2 ())
CHECK_VALID (true, RE, RE () | RE ())
CHECK_VALID (true, RE, RE () & RE ())
CHECK_VALID (true, RE, RE () ^ RE ())
/* operator OP (enum_flags, raw_enum) */
CHECK_VALID (false, void, EF () | RE2 ())
CHECK_VALID (false, void, EF () & RE2 ())
CHECK_VALID (false, void, EF () ^ RE2 ())
CHECK_VALID (true, EF, EF () | RE ())
CHECK_VALID (true, EF, EF () & RE ())
CHECK_VALID (true, EF, EF () ^ RE ())
/* operator OP= (raw_enum, raw_enum), rvalue ref on the lhs. */
CHECK_VALID (false, void, RE () |= RE2 ())
CHECK_VALID (false, void, RE () &= RE2 ())
CHECK_VALID (false, void, RE () ^= RE2 ())
CHECK_VALID (false, void, RE () |= RE ())
CHECK_VALID (false, void, RE () &= RE ())
CHECK_VALID (false, void, RE () ^= RE ())
/* operator OP= (raw_enum, raw_enum), lvalue ref on the lhs. */
CHECK_VALID (false, void, re |= RE2 ())
CHECK_VALID (false, void, re &= RE2 ())
CHECK_VALID (false, void, re ^= RE2 ())
CHECK_VALID (true, RE&, re |= RE ())
CHECK_VALID (true, RE&, re &= RE ())
CHECK_VALID (true, RE&, re ^= RE ())
/* operator OP= (enum_flags, raw_enum), rvalue ref on the lhs. */
CHECK_VALID (false, void, EF () |= RE2 ())
CHECK_VALID (false, void, EF () &= RE2 ())
CHECK_VALID (false, void, EF () ^= RE2 ())
CHECK_VALID (false, void, EF () |= RE ())
CHECK_VALID (false, void, EF () &= RE ())
CHECK_VALID (false, void, EF () ^= RE ())
/* operator OP= (enum_flags, raw_enum), lvalue ref on the lhs. */
CHECK_VALID (false, void, ef |= RE2 ())
CHECK_VALID (false, void, ef &= RE2 ())
CHECK_VALID (false, void, ef ^= RE2 ())
CHECK_VALID (true, EF&, ef |= EF ())
CHECK_VALID (true, EF&, ef &= EF ())
CHECK_VALID (true, EF&, ef ^= EF ())
/* operator OP= (enum_flags, enum_flags), rvalue ref on the lhs. */
CHECK_VALID (false, void, EF () |= EF2 ())
CHECK_VALID (false, void, EF () &= EF2 ())
CHECK_VALID (false, void, EF () ^= EF2 ())
CHECK_VALID (false, void, EF () |= EF ())
CHECK_VALID (false, void, EF () &= EF ())
CHECK_VALID (false, void, EF () ^= EF ())
/* operator OP= (enum_flags, enum_flags), lvalue ref on the lhs. */
CHECK_VALID (false, void, ef |= EF2 ())
CHECK_VALID (false, void, ef &= EF2 ())
CHECK_VALID (false, void, ef ^= EF2 ())
CHECK_VALID (true, EF&, ef |= EF ())
CHECK_VALID (true, EF&, ef &= EF ())
CHECK_VALID (true, EF&, ef ^= EF ())
/* operator~ (raw_enum) */
CHECK_VALID (false, void, ~RE ())
CHECK_VALID (true, URE, ~URE ())
/* operator~ (enum_flags) */
CHECK_VALID (false, void, ~EF ())
CHECK_VALID (true, UEF, ~UEF ())
/* Check ternary operator. This exercises implicit conversions. */
CHECK_VALID (true, EF, true ? EF () : RE ())
CHECK_VALID (true, EF, true ? RE () : EF ())
/* These are valid, but it's not a big deal since you won't be able to
assign the resulting integer to an enum or an enum_flags without a
cast.
The latter two tests are disabled on older GCCs because they
incorrectly fail with gcc 4.8 and 4.9 at least. Running the test
outside a SFINAE context shows:
invalid user-defined conversion from EF to RE2
They've been confirmed to compile/pass with gcc 5.3, gcc 7.1 and
clang 3.7. */
CHECK_VALID (true, int, true ? EF () : EF2 ())
CHECK_VALID (true, int, true ? EF2 () : EF ())
#if GCC_VERSION >= 5003 || defined __clang__
CHECK_VALID (true, int, true ? EF () : RE2 ())
CHECK_VALID (true, int, true ? RE2 () : EF ())
#endif
/* Same, but with an unsigned enum. */
typedef unsigned int uns;
CHECK_VALID (true, uns, true ? EF () : UEF ())
CHECK_VALID (true, uns, true ? UEF () : EF ())
#if GCC_VERSION >= 5003 || defined __clang__
CHECK_VALID (true, uns, true ? EF () : URE ())
CHECK_VALID (true, uns, true ? URE () : EF ())
#endif
/* Unfortunately this can't work due to the way C++ computes the
return type of the ternary conditional operator. int isn't
implicitly convertible to the raw enum type, so the type of the
expression is int. And then int is not implicitly convertible to
enum_flags.
GCC 4.8 fails to compile this test with:
error: operands to ?: have different types enum_flags<RE> and int
Confirmed to work with gcc 4.9, 5.3 and clang 3.7.
*/
#if GCC_VERSION >= 4009 || defined __clang__
CHECK_VALID (false, void, true ? EF () : 0)
CHECK_VALID (false, void, true ? 0 : EF ())
#endif
/* Check that the ++/--/<</>>/<<=/>>= operators are deleted. */
CHECK_VALID (false, void, RE ()++)
CHECK_VALID (false, void, ++RE ())
CHECK_VALID (false, void, --RE ())
CHECK_VALID (false, void, RE ()--)
CHECK_VALID (false, void, RE () << 1)
CHECK_VALID (false, void, RE () >> 1)
CHECK_VALID (false, void, EF () << 1)
CHECK_VALID (false, void, EF () >> 1)
CHECK_VALID (false, void, RE () <<= 1)
CHECK_VALID (false, void, RE () >>= 1)
CHECK_VALID (false, void, EF () <<= 1)
CHECK_VALID (false, void, EF () >>= 1)
/* Test comparison operators. */
CHECK_VALID (false, void, EF () == EF2 ())
CHECK_VALID (false, void, EF () == RE2 ())
CHECK_VALID (false, void, RE () == EF2 ())
CHECK_VALID (true, bool, EF (RE (1)) == EF (RE (1)))
CHECK_VALID (true, bool, EF (RE (1)) == RE (1))
CHECK_VALID (true, bool, RE (1) == EF (RE (1)))
CHECK_VALID (false, void, EF () != EF2 ())
CHECK_VALID (false, void, EF () != RE2 ())
CHECK_VALID (false, void, RE () != EF2 ())
Tweak gdbsupport/valid-expr.h for GCC 6, fix build With GCC 6.4 and 6.5 (at least), unit tests that use gdbsupport/valid-expr.h's CHECK_VALID fail to compile, with: In file included from src/gdb/unittests/offset-type-selftests.c:24:0: src/gdb/unittests/offset-type-selftests.c: In substitution of 'template<class Expected, template<class ...> class Op, class ... Args> using is_detected_exact = std::is_same<Expected, typename gdb::detection_detail::detector<gdb::nonesuch, void, Op, Args ...>::type> [with Expected = selftests::offset_type::off_A&; Op = selftests::offset_type::check_valid_expr75::archetype; Args = {selftests::offset_type::off_A, selftests::offset_type::off_B}]': src/gdb/unittests/offset-type-selftests.c:75:1: required from here src/gdb/../gdbsupport/valid-expr.h:65:20: error: type/value mismatch at argument 2 in template parameter list for 'template<class Expected, template<class ...> class Op, class ... Args> using is_detected_exact = std::is_same<Expected, typename gdb::detection_detail::detector<gdb::nonesuch, void, Op, Args ...>::type>' archetype, TYPES>::value == VALID, \ ^ The important part is the "error: type/value mismatch" error. Seems like that GCC doesn't understand that archetype is an alias template, and is being strict in requiring a template class. The fix here is then to make archetype a template class, to pacify GCC. The resulting code looks like this: template <TYPENAMES, typename = decltype (EXPR)> struct archetype { }; static_assert (gdb::is_detected_exact<archetype<TYPES, EXPR_TYPE>, archetype, TYPES>::value == VALID, ""); is_detected_exact<Expected, Op, Args> checks whether Op<Args> is type Expected: - For Expected, we pass the explicit EXPR_TYPE, overriding the default parameter type of archetype. - For Args we don't pass the last template parameter, so archtype defaults to the EXPR's decltype. So in essence, we're really checking whether EXPR_TYPE is the same as decltype(EXPR). We need to do the decltype in a template context in order to trigger SFINAE instead of failing to compile. The hunk in unittests/enum-flags-selftests.c becomes necessary, because unlike with the current alias template version, this new version makes GCC trigger -Wenum-compare warnings as well: src/gdb/unittests/enum-flags-selftests.c:328:33: error: comparison between 'enum selftests::enum_flags_tests::RE' and 'enum selftests::enum_flags_tests::RE2' [-Werror=enum-compare] CHECK_VALID (true, bool, RE () != RE2 ()) ^ src/gdb/../gdbsupport/valid-expr.h:61:45: note: in definition of macro 'CHECK_VALID_EXPR_INT' template <TYPENAMES, typename = decltype (EXPR)> \ ^ Build-tested with: - GCC {4.8.5, 6.4, 6.5, 7.3.1, 9.3.0, 11.0.0-20200910} - Clang 10.0.0 gdbsupport/ChangeLog: * valid-expr.h (CHECK_VALID_EXPR_INT): Make archetype a template class instead of an alias template and adjust static_assert. gdb/ChangeLog: * unittests/enum-flags-selftests.c: Check whether __GNUC__ is defined before using '#pragma GCC diagnostic' instead of checking __clang__.
2020-09-30 03:08:51 +08:00
/* Disable -Wenum-compare due to:
Clang:
"error: comparison of two values with different enumeration types
[-Werror,-Wenum-compare]"
GCC:
"error: comparison between enum selftests::enum_flags_tests::RE
and enum selftests::enum_flags_tests::RE2
[-Werror=enum-compare]"
Not a big deal since misuses like these in GDB will be caught by
-Werror anyway. This check is here mainly for completeness. */
#if defined __GNUC__
Rewrite enum_flags, add unit tests, fix problems This patch started by adding comprehensive unit tests for enum_flags. For the testing part, it adds: - tests of normal expected uses of the API. - checks that _invalid_ uses of the API would fail to compile. I.e., it validates that enum_flags really is a strong type, and that incorrect mixing of enum types would be caught at compile time. It pulls that off making use of SFINEA and C++11's decltype/constexpr. This revealed many holes in the enum_flags API. For example, the f1 assignment below currently incorrectly fails to compile: enum_flags<flags> f1 = FLAG1; enum_flags<flags> f2 = FLAG2 | f1; The unit tests also revealed that this useful use case doesn't work: enum flag { FLAG1 = 1, FLAG2 = 2 }; enum_flags<flag> src = FLAG1; enum_flags<flag> f1 = condition ? src : FLAG2; It fails to compile because enum_flags<flag> and flag are convertible to each other. Turns out that making enum_flags be implicitly convertible to the backing raw enum type was not a good idea. If we make it convertible to the underlying type instead, we fix that ternary operator use case, and, we find cases throughout the codebase that should be using the enum_flags but were using the raw backing enum instead. So it's a good change overall. Also, several operators were missing. These holes and more are plugged by this patch, by reworking how the enum_flags operators are implemented, and making use of C++11's feature of being able to delete methods/functions. There are cases in gdb/compile/ where we need to call a function in a C plugin API that expects the raw enum. To address cases like that, this adds a "raw()" method to enum_flags. This way we can keep using the safer enum_flags to construct the value, and then be explicit when we need to get at the raw enum. This makes most of the enum_flags operators constexpr. Beyond enabling more compiler optimizations and enabling the new unit tests, this has other advantages, like making it possible to use operator| with enum_flags values in switch cases, where only compile-time constants are allowed: enum_flags<flags> f = FLAG1 | FLAG2; switch (f) { case FLAG1 | FLAG2: break; } Currently that fails to compile. It also switches to a different mechanism of enabling the global operators. The current mechanism isn't namespace friendly, the new one is. It also switches to C++11-style SFINAE -- instead of wrapping the return type in a SFINAE-friently structure, we use an unnamed template parameter. I.e., this: template <typename enum_type, typename = is_enum_flags_enum_type_t<enum_type>> enum_type operator& (enum_type e1, enum_type e2) instead of: template <typename enum_type> typename enum_flags_type<enum_type>::type operator& (enum_type e1, enum_type e2) Note that the static_assert inside operator~() was converted to a couple overloads (signed vs unsigned), because static_assert is too late for SFINAE-based tests, which is important for the CHECK_VALID unit tests. Tested with gcc {4.8, 7.1, 9.3} and clang {5.0.2, 10.0.0}. gdb/ChangeLog: * Makefile.in (SELFTESTS_SRCS): Add unittests/enum-flags-selftests.c. * btrace.c (ftrace_update_caller, ftrace_fixup_calle): Use btrace_function_flags instead of enum btrace_function_flag. * compile/compile-c-types.c (convert_qualified): Use enum_flags::raw. * compile/compile-cplus-symbols.c (convert_one_symbol) (convert_symbol_bmsym): * compile/compile-cplus-types.c (compile_cplus_convert_method) (compile_cplus_convert_struct_or_union_methods) (compile_cplus_instance::convert_qualified_base): * go-exp.y (parse_string_or_char): Add cast to int. * unittests/enum-flags-selftests.c: New file. * record-btrace.c (btrace_thread_flag_to_str): Change parameter's type to btrace_thread_flags from btrace_thread_flag. (record_btrace_cancel_resume, record_btrace_step_thread): Change local's type to btrace_thread_flags from btrace_thread_flag. Add cast in DEBUG call. gdbsupport/ChangeLog: * enum-flags.h: Include "traits.h". (DEF_ENUM_FLAGS_TYPE): Declare a function instead of defining a structure. (enum_underlying_type): Update comment. (namespace enum_flags_detail): New. Move struct zero_type here. (EnumIsUnsigned, EnumIsSigned): New. (class enum_flags): Make most methods constexpr. (operator&=, operator|=, operator^=): Take an enum_flags instead of an enum_type. Make rvalue ref versions deleted. (operator enum_type()): Delete. (operator&, operator|, operator^, operator~): Delete, moved out of class. (raw()): New method. (is_enum_flags_enum_type_t): Declare. (ENUM_FLAGS_GEN_BINOP, ENUM_FLAGS_GEN_COMPOUND_ASSIGN) (ENUM_FLAGS_GEN_COMP): New. Use them to reimplement global operators. (operator~): Now constexpr and reimplemented. (operator<<, operator>>): New deleted functions. * valid-expr.h (CHECK_VALID_EXPR_5, CHECK_VALID_EXPR_6): New.
2020-09-15 04:16:59 +08:00
# pragma GCC diagnostic push
# pragma GCC diagnostic ignored "-Wenum-compare"
#endif
CHECK_VALID (true, bool, RE () == RE2 ())
CHECK_VALID (true, bool, RE () != RE2 ())
Tweak gdbsupport/valid-expr.h for GCC 6, fix build With GCC 6.4 and 6.5 (at least), unit tests that use gdbsupport/valid-expr.h's CHECK_VALID fail to compile, with: In file included from src/gdb/unittests/offset-type-selftests.c:24:0: src/gdb/unittests/offset-type-selftests.c: In substitution of 'template<class Expected, template<class ...> class Op, class ... Args> using is_detected_exact = std::is_same<Expected, typename gdb::detection_detail::detector<gdb::nonesuch, void, Op, Args ...>::type> [with Expected = selftests::offset_type::off_A&; Op = selftests::offset_type::check_valid_expr75::archetype; Args = {selftests::offset_type::off_A, selftests::offset_type::off_B}]': src/gdb/unittests/offset-type-selftests.c:75:1: required from here src/gdb/../gdbsupport/valid-expr.h:65:20: error: type/value mismatch at argument 2 in template parameter list for 'template<class Expected, template<class ...> class Op, class ... Args> using is_detected_exact = std::is_same<Expected, typename gdb::detection_detail::detector<gdb::nonesuch, void, Op, Args ...>::type>' archetype, TYPES>::value == VALID, \ ^ The important part is the "error: type/value mismatch" error. Seems like that GCC doesn't understand that archetype is an alias template, and is being strict in requiring a template class. The fix here is then to make archetype a template class, to pacify GCC. The resulting code looks like this: template <TYPENAMES, typename = decltype (EXPR)> struct archetype { }; static_assert (gdb::is_detected_exact<archetype<TYPES, EXPR_TYPE>, archetype, TYPES>::value == VALID, ""); is_detected_exact<Expected, Op, Args> checks whether Op<Args> is type Expected: - For Expected, we pass the explicit EXPR_TYPE, overriding the default parameter type of archetype. - For Args we don't pass the last template parameter, so archtype defaults to the EXPR's decltype. So in essence, we're really checking whether EXPR_TYPE is the same as decltype(EXPR). We need to do the decltype in a template context in order to trigger SFINAE instead of failing to compile. The hunk in unittests/enum-flags-selftests.c becomes necessary, because unlike with the current alias template version, this new version makes GCC trigger -Wenum-compare warnings as well: src/gdb/unittests/enum-flags-selftests.c:328:33: error: comparison between 'enum selftests::enum_flags_tests::RE' and 'enum selftests::enum_flags_tests::RE2' [-Werror=enum-compare] CHECK_VALID (true, bool, RE () != RE2 ()) ^ src/gdb/../gdbsupport/valid-expr.h:61:45: note: in definition of macro 'CHECK_VALID_EXPR_INT' template <TYPENAMES, typename = decltype (EXPR)> \ ^ Build-tested with: - GCC {4.8.5, 6.4, 6.5, 7.3.1, 9.3.0, 11.0.0-20200910} - Clang 10.0.0 gdbsupport/ChangeLog: * valid-expr.h (CHECK_VALID_EXPR_INT): Make archetype a template class instead of an alias template and adjust static_assert. gdb/ChangeLog: * unittests/enum-flags-selftests.c: Check whether __GNUC__ is defined before using '#pragma GCC diagnostic' instead of checking __clang__.
2020-09-30 03:08:51 +08:00
#if defined __GNUC__
Rewrite enum_flags, add unit tests, fix problems This patch started by adding comprehensive unit tests for enum_flags. For the testing part, it adds: - tests of normal expected uses of the API. - checks that _invalid_ uses of the API would fail to compile. I.e., it validates that enum_flags really is a strong type, and that incorrect mixing of enum types would be caught at compile time. It pulls that off making use of SFINEA and C++11's decltype/constexpr. This revealed many holes in the enum_flags API. For example, the f1 assignment below currently incorrectly fails to compile: enum_flags<flags> f1 = FLAG1; enum_flags<flags> f2 = FLAG2 | f1; The unit tests also revealed that this useful use case doesn't work: enum flag { FLAG1 = 1, FLAG2 = 2 }; enum_flags<flag> src = FLAG1; enum_flags<flag> f1 = condition ? src : FLAG2; It fails to compile because enum_flags<flag> and flag are convertible to each other. Turns out that making enum_flags be implicitly convertible to the backing raw enum type was not a good idea. If we make it convertible to the underlying type instead, we fix that ternary operator use case, and, we find cases throughout the codebase that should be using the enum_flags but were using the raw backing enum instead. So it's a good change overall. Also, several operators were missing. These holes and more are plugged by this patch, by reworking how the enum_flags operators are implemented, and making use of C++11's feature of being able to delete methods/functions. There are cases in gdb/compile/ where we need to call a function in a C plugin API that expects the raw enum. To address cases like that, this adds a "raw()" method to enum_flags. This way we can keep using the safer enum_flags to construct the value, and then be explicit when we need to get at the raw enum. This makes most of the enum_flags operators constexpr. Beyond enabling more compiler optimizations and enabling the new unit tests, this has other advantages, like making it possible to use operator| with enum_flags values in switch cases, where only compile-time constants are allowed: enum_flags<flags> f = FLAG1 | FLAG2; switch (f) { case FLAG1 | FLAG2: break; } Currently that fails to compile. It also switches to a different mechanism of enabling the global operators. The current mechanism isn't namespace friendly, the new one is. It also switches to C++11-style SFINAE -- instead of wrapping the return type in a SFINAE-friently structure, we use an unnamed template parameter. I.e., this: template <typename enum_type, typename = is_enum_flags_enum_type_t<enum_type>> enum_type operator& (enum_type e1, enum_type e2) instead of: template <typename enum_type> typename enum_flags_type<enum_type>::type operator& (enum_type e1, enum_type e2) Note that the static_assert inside operator~() was converted to a couple overloads (signed vs unsigned), because static_assert is too late for SFINAE-based tests, which is important for the CHECK_VALID unit tests. Tested with gcc {4.8, 7.1, 9.3} and clang {5.0.2, 10.0.0}. gdb/ChangeLog: * Makefile.in (SELFTESTS_SRCS): Add unittests/enum-flags-selftests.c. * btrace.c (ftrace_update_caller, ftrace_fixup_calle): Use btrace_function_flags instead of enum btrace_function_flag. * compile/compile-c-types.c (convert_qualified): Use enum_flags::raw. * compile/compile-cplus-symbols.c (convert_one_symbol) (convert_symbol_bmsym): * compile/compile-cplus-types.c (compile_cplus_convert_method) (compile_cplus_convert_struct_or_union_methods) (compile_cplus_instance::convert_qualified_base): * go-exp.y (parse_string_or_char): Add cast to int. * unittests/enum-flags-selftests.c: New file. * record-btrace.c (btrace_thread_flag_to_str): Change parameter's type to btrace_thread_flags from btrace_thread_flag. (record_btrace_cancel_resume, record_btrace_step_thread): Change local's type to btrace_thread_flags from btrace_thread_flag. Add cast in DEBUG call. gdbsupport/ChangeLog: * enum-flags.h: Include "traits.h". (DEF_ENUM_FLAGS_TYPE): Declare a function instead of defining a structure. (enum_underlying_type): Update comment. (namespace enum_flags_detail): New. Move struct zero_type here. (EnumIsUnsigned, EnumIsSigned): New. (class enum_flags): Make most methods constexpr. (operator&=, operator|=, operator^=): Take an enum_flags instead of an enum_type. Make rvalue ref versions deleted. (operator enum_type()): Delete. (operator&, operator|, operator^, operator~): Delete, moved out of class. (raw()): New method. (is_enum_flags_enum_type_t): Declare. (ENUM_FLAGS_GEN_BINOP, ENUM_FLAGS_GEN_COMPOUND_ASSIGN) (ENUM_FLAGS_GEN_COMP): New. Use them to reimplement global operators. (operator~): Now constexpr and reimplemented. (operator<<, operator>>): New deleted functions. * valid-expr.h (CHECK_VALID_EXPR_5, CHECK_VALID_EXPR_6): New.
2020-09-15 04:16:59 +08:00
# pragma GCC diagnostic pop
#endif
CHECK_VALID (true, bool, EF (RE (1)) != EF (RE (2)))
CHECK_VALID (true, bool, EF (RE (1)) != RE (2))
CHECK_VALID (true, bool, RE (1) != EF (RE (2)))
CHECK_VALID (true, bool, EF () == 0)
/* Check we didn't disable/delete comparison between non-flags enums
and unrelated types by mistake. */
CHECK_VALID (true, bool, NF (1) == NF (1))
CHECK_VALID (true, bool, NF (1) == int (1))
CHECK_VALID (true, bool, NF (1) == char (1))
/* -------------------------------------------------------------------- */
/* Follows misc tests that exercise the API. Some are compile time,
when possible, others are run time. */
enum test_flag
{
FLAG1 = 1 << 1,
FLAG2 = 1 << 2,
FLAG3 = 1 << 3,
};
enum test_uflag : unsigned
{
UFLAG1 = 1 << 1,
UFLAG2 = 1 << 2,
UFLAG3 = 1 << 3,
};
DEF_ENUM_FLAGS_TYPE (test_flag, test_flags);
DEF_ENUM_FLAGS_TYPE (test_uflag, test_uflags);
static void
self_test ()
{
/* Check that default construction works. */
{
constexpr test_flags f;
gdb_static_assert (f == 0);
}
/* Check that assignment from zero works. */
{
test_flags f (FLAG1);
SELF_CHECK (f == FLAG1);
f = 0;
SELF_CHECK (f == 0);
}
/* Check that construction from zero works. */
{
constexpr test_flags zero1 = 0;
constexpr test_flags zero2 (0);
constexpr test_flags zero3 {0};
constexpr test_flags zero4 = {0};
gdb_static_assert (zero1 == 0);
gdb_static_assert (zero2 == 0);
gdb_static_assert (zero3 == 0);
gdb_static_assert (zero4 == 0);
}
/* Check construction from enum value. */
{
gdb_static_assert (test_flags (FLAG1) == FLAG1);
gdb_static_assert (test_flags (FLAG2) != FLAG1);
}
/* Check copy/assignment. */
{
constexpr test_flags src = FLAG1;
constexpr test_flags f1 = src;
constexpr test_flags f2 (src);
constexpr test_flags f3 {src};
constexpr test_flags f4 = {src};
gdb_static_assert (f1 == FLAG1);
gdb_static_assert (f2 == FLAG1);
gdb_static_assert (f3 == FLAG1);
gdb_static_assert (f4 == FLAG1);
}
/* Check moving. */
{
test_flags src = FLAG1;
test_flags dst = 0;
dst = std::move (src);
SELF_CHECK (dst == FLAG1);
}
/* Check construction from an 'or' of multiple bits. For this to
work, operator| must be overridden to return an enum type. The
builtin version would return int instead and then the conversion
to test_flags would fail. */
{
constexpr test_flags f = FLAG1 | FLAG2;
gdb_static_assert (f == (FLAG1 | FLAG2));
}
/* Similarly, check that "FLAG1 | FLAG2" on the rhs of an assignment
operator works. */
{
test_flags f = 0;
f |= FLAG1 | FLAG2;
SELF_CHECK (f == (FLAG1 | FLAG2));
f &= FLAG1 | FLAG2;
SELF_CHECK (f == (FLAG1 | FLAG2));
f ^= FLAG1 | FLAG2;
SELF_CHECK (f == 0);
}
/* Check explicit conversion to int works. */
{
constexpr int some_bits (FLAG1 | FLAG2);
/* And comparison with int works too. */
gdb_static_assert (some_bits == (FLAG1 | FLAG2));
gdb_static_assert (some_bits == test_flags (FLAG1 | FLAG2));
}
/* Check operator| and operator|=. Particularly interesting is
making sure that putting the enum value on the lhs side of the
expression works (FLAG | f). */
{
test_flags f = FLAG1;
f |= FLAG2;
SELF_CHECK (f == (FLAG1 | FLAG2));
}
{
test_flags f = FLAG1;
f = f | FLAG2;
SELF_CHECK (f == (FLAG1 | FLAG2));
}
{
test_flags f = FLAG1;
f = FLAG2 | f;
SELF_CHECK (f == (FLAG1 | FLAG2));
}
/* Check the &/&= operators. */
{
test_flags f = FLAG1 & FLAG2;
SELF_CHECK (f == 0);
f = FLAG1 | FLAG2;
f &= FLAG2;
SELF_CHECK (f == FLAG2);
f = FLAG1 | FLAG2;
f = f & FLAG2;
SELF_CHECK (f == FLAG2);
f = FLAG1 | FLAG2;
f = FLAG2 & f;
SELF_CHECK (f == FLAG2);
}
/* Check the ^/^= operators. */
{
constexpr test_flags f = FLAG1 ^ FLAG2;
gdb_static_assert (f == (FLAG1 ^ FLAG2));
}
{
test_flags f = FLAG1 ^ FLAG2;
f ^= FLAG3;
SELF_CHECK (f == (FLAG1 | FLAG2 | FLAG3));
f = f ^ FLAG3;
SELF_CHECK (f == (FLAG1 | FLAG2));
f = FLAG3 ^ f;
SELF_CHECK (f == (FLAG1 | FLAG2 | FLAG3));
}
/* Check operator~. Note this only compiles with unsigned
flags. */
{
constexpr test_uflags f1 = ~UFLAG1;
constexpr test_uflags f2 = ~f1;
gdb_static_assert (f2 == UFLAG1);
}
/* Check the ternary operator. */
{
/* raw enum, raw enum */
constexpr test_flags f1 = true ? FLAG1 : FLAG2;
gdb_static_assert (f1 == FLAG1);
constexpr test_flags f2 = false ? FLAG1 : FLAG2;
gdb_static_assert (f2 == FLAG2);
}
{
/* enum flags, raw enum */
constexpr test_flags src = FLAG1;
constexpr test_flags f1 = true ? src : FLAG2;
gdb_static_assert (f1 == FLAG1);
constexpr test_flags f2 = false ? src : FLAG2;
gdb_static_assert (f2 == FLAG2);
}
{
/* enum flags, enum flags */
constexpr test_flags src1 = FLAG1;
constexpr test_flags src2 = FLAG2;
constexpr test_flags f1 = true ? src1 : src2;
gdb_static_assert (f1 == src1);
constexpr test_flags f2 = false ? src1 : src2;
gdb_static_assert (f2 == src2);
}
/* Check that we can use flags in switch expressions (requires
unambiguous conversion to integer). Also check that we can use
operator| in switch cases, where only constants are allowed.
This should work because operator| is constexpr. */
{
test_flags f = FLAG1 | FLAG2;
bool ok = false;
switch (f)
{
case FLAG1:
break;
case FLAG2:
break;
case FLAG1 | FLAG2:
ok = true;
break;
}
SELF_CHECK (ok);
}
}
} /* namespace enum_flags_tests */
} /* namespace selftests */
void _initialize_enum_flags_selftests ();
void
_initialize_enum_flags_selftests ()
{
selftests::register_test ("enum-flags",
selftests::enum_flags_tests::self_test);
}