This is a major change on how we represent nested name qualifications in
the AST.
* The nested name specifier itself and how it's stored is changed. The
prefixes for types are handled within the type hierarchy, which makes
canonicalization for them super cheap, no memory allocation required.
Also translating a type into nested name specifier form becomes a no-op.
An identifier is stored as a DependentNameType. The nested name
specifier gains a lightweight handle class, to be used instead of
passing around pointers, which is similar to what is implemented for
TemplateName. There is still one free bit available, and this handle can
be used within a PointerUnion and PointerIntPair, which should keep
bit-packing aficionados happy.
* The ElaboratedType node is removed, all type nodes in which it could
previously apply to can now store the elaborated keyword and name
qualifier, tail allocating when present.
* TagTypes can now point to the exact declaration found when producing
these, as opposed to the previous situation of there only existing one
TagType per entity. This increases the amount of type sugar retained,
and can have several applications, for example in tracking module
ownership, and other tools which care about source file origins, such as
IWYU. These TagTypes are lazily allocated, in order to limit the
increase in AST size.
This patch offers a great performance benefit.
It greatly improves compilation time for
[stdexec](https://github.com/NVIDIA/stdexec). For one datapoint, for
`test_on2.cpp` in that project, which is the slowest compiling test,
this patch improves `-c` compilation time by about 7.2%, with the
`-fsyntax-only` improvement being at ~12%.
This has great results on compile-time-tracker as well:
This patch also further enables other optimziations in the future, and
will reduce the performance impact of template specialization resugaring
when that lands.
It has some other miscelaneous drive-by fixes.
About the review: Yes the patch is huge, sorry about that. Part of the
reason is that I started by the nested name specifier part, before the
ElaboratedType part, but that had a huge performance downside, as
ElaboratedType is a big performance hog. I didn't have the steam to go
back and change the patch after the fact.
There is also a lot of internal API changes, and it made sense to remove
ElaboratedType in one go, versus removing it from one type at a time, as
that would present much more churn to the users. Also, the nested name
specifier having a different API avoids missing changes related to how
prefixes work now, which could make existing code compile but not work.
How to review: The important changes are all in
`clang/include/clang/AST` and `clang/lib/AST`, with also important
changes in `clang/lib/Sema/TreeTransform.h`.
The rest and bulk of the changes are mostly consequences of the changes
in API.
PS: TagType::getDecl is renamed to `getOriginalDecl` in this patch, just
for easier to rebasing. I plan to rename it back after this lands.
Fixes#136624
Fixes https://github.com/llvm/llvm-project/issues/43179
Fixes https://github.com/llvm/llvm-project/issues/68670
Fixes https://github.com/llvm/llvm-project/issues/92757
Currently, the library-internal feature test macros are only defined if
the feature is not available, and always have the prefix
`_LIBCPP_HAS_NO_`. This patch changes that, so that they are always
defined and have the prefix `_LIBCPP_HAS_` instead. This changes the
canonical use of these macros to `#if _LIBCPP_HAS_FEATURE`, which means
that using an undefined macro (e.g. due to a missing include) is
diagnosed now. While this is rather unlikely currently, a similar change
in `<__configuration/availability.h>` caught a few bugs. This also
improves readability, since it removes the double-negation of `#ifndef
_LIBCPP_HAS_NO_FEATURE`.
The current patch only touches the macros defined in `<__config>`. If
people are happy with this approach, I'll make a follow-up PR to also
change the macros defined in `<__config_site>`.
This patch adds a large number of missing includes in the libc++ headers
and the test suite. Those were found as part of the effort to move
towards a mostly monolithic top-level std module.
This PR is a followup to #81080.
This PR makes two major changes to how the LCG operation is computed:
The first is that I added an additional case where `ax + c` might
overflow the intermediate variable, but `ax` by itself won't. In this
case, it's much better to use `(ax mod m) + c mod m` than the previous
behavior of falling back to Schrage's algorithm. The addition modulo is
done in the same way as when using Schrage's algorithm (i.e. `x += c -
(x >= m - c)*m`), but the multiplication modulo is calculated directly,
which is faster.
The second is that I added handling for the case where the `ax`
intermediate might overflow, but Schrage's algorithm doesn't apply (i.e.
r > q). In this case, the only real option is to increase the precision
of the intermediate values. The good news is that - for `x`, `a`, and
`c` being n-bit values - `ax + c` will never overflow a 2n-bit
intermediary, meaning this promotion can only happen once, and will
always be able to use the simplest implementation. This is already the
case for 16-bit LCGs, as libcxx chooses to compute them with 32-bit
intermediate values. For 32-bit LCGs, I simply added code similar to the
16-bit case to use the existing 64-bit implementations. Lastly, for
64-bit LCGs, I wrote a case that calculates it using `unsigned __int128`
if it is available to use.
While this implementation covers a *lot* of the missing cases from
#81080, this still won't compile **every** possible
`linear_congruential_engine`. Specifically, if `a`, `c`, and `m` are
chosen such that it needs 128-bit integers, but the platform doesn't
support `__int128` (eg. 32-bit x86), then it will fail to compile.
However, this is a fairly rare case to see actually used, and libcxx
would be in good company with this, as [libstdc++ also fails to compile
under these
circumstances](https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87744).
Fixing **this** gap would require even **more** work of further
complexity, so that would probably be best handled by a different PR
(I'll put more details on what that PR would entail in a comment).
This fixes two major mistakes in the implementation of
`linear_congruential_engine` that allowed it to produce incorrect
output. Specifically, these mistakes are in `__lce_alg_picker`, which is
used to determine whether Schrage's algorithm is valid and needed.
The first mistake is in the definition of `_OverflowOK`. The code
comment and the description of [D65041](https://reviews.llvm.org/D65041)
both indicate that it's supposed to be true iff `m` is a power of two.
However, the definition used does not work out to that, and instead is
true whenever `m` is even. This could result in
`linear_congruential_engine` using an invalid implementation, as it
would incorrectly assume that any integer overflow can't change the
result. I changed the implementation to one that accurately checks if
`m` is a power of two. Technically, this implementation has an edge case
where it considers `0` to be a power of two, but in this case this is
actually accurate behavior, as `m = 0` indicates a modulus of 2^w where
w is the size of `result_type` in bits, which *is* a power of two.
The second mistake is in the static assert. The original static assert
erroneously included an unnecessary `a != 0 || m != 0`. Combined with
the `|| !_MightOverflow`, this actually resulted in the static assert
being impossible to fail. Applying De Morgan's law and expanding
`_MightOverflow` gives that the only way this static assert can be
triggered is if `a == 0 && m == 0 && a != 0 && m != 0 && ...`, which
clearly cannot be true. I simply removed the explicit checks against `a`
and `m`, as the intended checks are already included in `_MightOverflow`
and `_SchrageOK`, and their inclusion doesn't provide any obvious
semantic benefit.
This should fix all the current instances where
`linear_congruential_engine` uses an invalid implementation. This
technically isn't a complete implementation, though, since the static
assert will cause some instantiations of `linear_congruential_engine`
not disallowed by the standard from compiling. However, this should
still be an improvement, as all compiling instantiations of
`linear_congruential_engine` should use a valid implementation. Fixing
the cases where the static assert triggers will require adding
additional implementations, some of which will be fairly non-trivial, so
I'd rather leave those for another PR so they don't hold up these more
important fixes.
Fixes#33554
I made sure they all had some expected-error output in them. Many of
these tests would be better implemented as a positive test using SFINAE,
but this is beyond the scope of this patch.
Differential Revision: https://reviews.llvm.org/D153980
This has been done using the following command
find libcxx/test -type f -exec perl -pi -e 's|^([^/]+?)((?<!::)(?<!::u)u?int(_[a-z]+)?[0-9]{1,2}_t)|\1std::\2|' \{} \;
And manually removed some false positives in std/depr/depr.c.headers.
Reviewed By: ldionne, #libc
Differential Revision: https://reviews.llvm.org/D145880
This patch rewords the static assert diagnostic output. Failing a
_Static_assert in C should not report that static_assert failed. This
changes the wording to be more like GCC and uses "static assertion"
when possible instead of hard coding the name. This also changes some
instances of 'static_assert' to instead be based on the token in the
source code.
Differential Revision: https://reviews.llvm.org/D129048
Looks like we again are going to have problems with libcxx tests that
are overly specific in their dependency on clang's diagnostics.
This reverts commit 6542cb55a3eb115b1c3592514590a19987ffc498.
This patch is basically the rewording of the static assert statement's
output(error) on screen after failing. Failing a _Static_assert in C
should not report that static_assert failed. It’d probably be better to
reword the diagnostic to be more like GCC and say “static assertion”
failed in both C and C++.
consider a c file having code
_Static_assert(0, "oh no!");
In clang the output is like:
<source>:1:1: error: static_assert failed: oh no!
_Static_assert(0, "oh no!");
^ ~
1 error generated.
Compiler returned: 1
Thus here the "static_assert" is not much good, it will be better to
reword it to the "static assertion failed" to more generic. as the gcc
prints as:
<source>:1:1: error: static assertion failed: "oh no!"
1 | _Static_assert(0, "oh no!");
| ^~~~~~~~~~~~~~
Compiler returned: 1
The above can also be seen here. This patch is about rewording
the static_assert to static assertion.
Differential Revision: https://reviews.llvm.org/D129048
Display 'static_assert failed: message' instead of
'static_assert failed "message"' to be consistent
with other implementations and be slightly more
readable.
Reviewed By: #libc, aaron.ballman, philnik, Mordante
Differential Revision: https://reviews.llvm.org/D128844
Since those features are general properties of the environment, it makes
sense to use them from libc++abi too, and so the name libcpp-has-no-xxx
doesn't make sense.
Differential Revision: https://reviews.llvm.org/D126482
... it's easier to suppress warnings internally, where we can detect the compiler.
* Rename `TEST_COMPILER_C1XX` to `TEST_COMPILER_MSVC`
* Rename all `TEST_WORKAROUND_C1XX_<meow>` to `TEST_WORKAROUND_MSVC_<meow>`
Differential Revision: https://reviews.llvm.org/D117422
Actually there's one functional change here, which is that users can
no longer depend on <random> to include all of C++20 <concepts>. That
inclusion is so new that we believe nobody should be depending on it
yet, even in the presence of Hyrum's Law. We keep the includes of <vector>,
<algorithm>, etc., so as not to break pre-C++20 Hyrum's Law users.
Differential Revision: https://reviews.llvm.org/D114281
This will avoid typos like `_LIBCPP_STD_VERS` (<future>) or using `#if TEST_STD_VER > 17` without including "test_macros.h".
Reviewed By: ldionne, #libc
Differential Revision: https://reviews.llvm.org/D99515
This patch changes how linear_congruential_engine picks its randomization
algorithm. It adds two restrictions, `_OverflowOK` and `_SchrageOK`.
`_OverflowOK` means that m is a power of two so using the classic
`(a * x + c) % m` will create a meaningless overflow. The second checks
that Schrage's algorithm will produce results that are in bounds of min
and max. This patch fixes https://llvm.org/PR27839.
Differential Revision: D65041
When porting libc++ to embedded systems, it can be useful to drop support
for localization, which these systems don't implement or care about.
Differential Revision: https://reviews.llvm.org/D90072
The testing script used to test libc++ historically did not like directories
without any testing files, so these tests had been added. Since this is
not necessary anymore, we can now remove these files. This has the benefit
that the total number of tests reflects the real number of tests more
closely, and we also skip some unnecessary work (especially relevant when
running tests over SSH).
However, some nothing_to_do.pass.cpp tests actually serve the purpose of
documenting that an area of the Standard doesn't need to be tested, or is
tested elsewhere. These files are not removed by this commit.
Removal done with:
import os
import itertools
for (dirpath, dirnames, filenames) in itertools.chain(os.walk('./libcxx/test'),
os.walk('./libcxxabi/test')):
if len(filenames + dirnames) > 1 and \
any(p == 'nothing_to_do.pass.cpp' for p in filenames):
os.remove(os.path.join(dirpath, 'nothing_to_do.pass.cpp'))
Summary:
Freestanding is *weird*. The standard allows it to differ in a bunch of odd
manners from regular C++, and the committee would like to improve that
situation. I'd like to make libc++ behave better with what freestanding should
be, so that it can be a tool we use in improving the standard. To do that we
need to try stuff out, both with "freestanding the language mode" and
"freestanding the library subset".
Let's start with the super basic: run the libc++ tests in freestanding, using
clang as the compiler, and see what works. The easiest hack to do this:
In utils/libcxx/test/config.py add:
self.cxx.compile_flags += ['-ffreestanding']
Run the tests and they all fail.
Why? Because in freestanding `main` isn't special. This "not special" property
has two effects: main doesn't get mangled, and main isn't allowed to omit its
`return` statement. The first means main gets mangled and the linker can't
create a valid executable for us to test. The second means we spew out warnings
(ew) and the compiler doesn't insert the `return` we omitted, and main just
falls of the end and does whatever undefined behavior (if you're luck, ud2
leading to non-zero return code).
Let's start my work with the basics. This patch changes all libc++ tests to
declare `main` as `int main(int, char**` so it mangles consistently (enabling us
to declare another `extern "C"` main for freestanding which calls the mangled
one), and adds `return 0;` to all places where it was missing. This touches 6124
files, and I apologize.
The former was done with The Magic Of Sed.
The later was done with a (not quite correct but decent) clang tool:
https://gist.github.com/jfbastien/793819ff360baa845483dde81170feed
This works for most tests, though I did have to adjust a few places when e.g.
the test runs with `-x c`, macros are used for main (such as for the filesystem
tests), etc.
Once this is in we can create a freestanding bot which will prevent further
regressions. After that, we can start the real work of supporting C++
freestanding fairly well in libc++.
<rdar://problem/47754795>
Reviewers: ldionne, mclow.lists, EricWF
Subscribers: christof, jkorous, dexonsmith, arphaman, miyuki, libcxx-commits
Differential Revision: https://reviews.llvm.org/D57624
llvm-svn: 353086
to reflect the new license. These used slightly different spellings that
defeated my regular expressions.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351648
Summary:
When a seed sequence would lead to having no non-zero significant bits
in the initial state of a `mersenne_twister_engine`, the fallback is to
flip the most significant bit of the first value that appears in the
textual representation of the initial state.
rand.eng.mers describes this as setting the value to be 2 to the power
of one less than w; the previous value encoded in the implementation,
namely one less than "2 to the power of w", is replaced by the correct
value in this patch.
Reviewers: mclow.lists, EricWF, jasonliu
Reviewed By: mclow.lists
Subscribers: mclow.lists, jasonliu, EricWF, christof, ldionne, cfe-commits
Differential Revision: https://reviews.llvm.org/D50736
llvm-svn: 339969
test/std/input.output/iostream.format/input.streams/istream.unformatted/get.pass.cpp
Add static_cast<char> because basic_istream::get() returns int_type (N4606 27.7.2.3 [istream.unformatted]/4).
test/std/input.output/iostream.format/output.streams/ostream.formatted/ostream.inserters.arithmetic/minus1.pass.cpp
Add static_cast<char> because toupper() returns int (C11 7.4.2.2/1).
test/std/iterators/stream.iterators/ostream.iterator/ostream.iterator.ops/assign_t.pass.cpp
This test is intentionally writing doubles to ostream_iterator<int>.
It's silencing -Wliteral-conversion for Clang, so I'm adding C4244 silencing for MSVC.
test/std/language.support/support.limits/limits/numeric.limits.members/infinity.pass.cpp
Given `extern float zero;`, the expression `1./zero` has type double, which emits a truncation warning
when being passed to test<float>() taking float. The fix is to say `1.f/zero` which has type float.
test/std/numerics/complex.number/cmplx.over/arg.pass.cpp
test/std/numerics/complex.number/cmplx.over/norm.pass.cpp
These tests were constructing std::complex<double>(x, 0), emitting truncation warnings when x is long long.
Saying static_cast<double>(x) avoids this.
test/std/numerics/rand/rand.eng/rand.eng.lcong/seed_result_type.pass.cpp
This was using `int s` to construct and seed a linear_congruential_engine<T, stuff>, where T is
unsigned short/unsigned int/unsigned long/unsigned long long. That emits a truncation warning in the
unsigned short case. Because the range [0, 20) is tiny and we aren't doing anything else with the index,
we can just iterate with `T s`.
test/std/re/re.traits/value.pass.cpp
regex_traits<wchar_t>::value()'s first parameter is wchar_t (N4606 28.7 [re.traits]/13). This loop is
using int to iterate through ['g', 0xFFFF), emitting a truncation warning from int to wchar_t
(which is 16-bit for some of us). Because the bound is exclusive, we can just iterate with wchar_t.
test/std/strings/basic.string/string.cons/size_char_alloc.pass.cpp
This test is a little strange. It's trying to verify that basic_string's (InIt, InIt) range constructor
isn't confused by "N copies of C" when N and C have the same integral type. To do this, it was
testing (100, 65), but that eventually emits truncation warnings from int to char. There's a simple way
to avoid this - passing (static_cast<char>(100), static_cast<char>(65)) also exercises the disambiguation.
(And 100 is representable even when char has a signed range.)
test/std/strings/string.view/string.view.hash/string_view.pass.cpp
Add static_cast<char_type> because `'0' + i` has type int.
test/std/utilities/function.objects/bind/func.bind/func.bind.bind/nested.pass.cpp
What's more horrible than nested bind()? pow() overloads! This operator()(T a, T b) was assuming that
std::pow(a, b) can be returned as T. (In this case, T is int.) However, N4606 26.9.1 [cmath.syn]/2
says that pow(int, int) returns double, so this was truncating double to int.
Adding static_cast<T> silences this.
test/std/utilities/function.objects/unord.hash/integral.pass.cpp
This was iterating `for (int i = 0; i <= 5; ++i)` and constructing `T t(i);` but that's truncating
when T is short. (And super truncating when T is bool.) Adding static_cast<T> silences this.
test/std/utilities/utility/exchange/exchange.pass.cpp
First, this was exchanging 67.2 into an int, but that's inherently truncating.
Changing this to static_cast<short>(67) avoids the truncation while preserving the
"what if T and U are different" test coverage.
Second, this was exchanging {} with the explicit type float into an int, and that's also
inherently truncating. Specifying short is just as good.
test/std/utilities/utility/pairs/pairs.spec/make_pair.pass.cpp
Add static_cast<short>. Note that this affects template argument deduction for make_pair(),
better fulfilling the test's intent. For example, this was saying
`typedef std::pair<int, short> P1; P1 p1 = std::make_pair(3, 4);` but that was asking
make_pair() to return pair<int, int>, which was then being converted to pair<int, short>.
(pair's converting constructors are tested elsewhere.)
Now, std::make_pair(3, static_cast<short>(4)) actually returns pair<int, short>.
(There's still a conversion from pair<nullptr_t, short> to pair<unique_ptr<int>, short>.)
Fixes D27544.
llvm-svn: 289111
When initializing unsigned integers to their maximum values, change "const T M(~0);" to "const T M(static_cast<T>(-1));".
~0 and -1 are equivalent, but I consider the -1 form to be significantly clearer (and more consistent with other tests).
llvm-svn: 287827
In C++11 mode and newer, use real static_asserts.
In C++03 mode, min() and max() aren't constexpr, so use plain asserts.
One test triggers MSVC's warning C4310 "cast truncates constant value".
The code is valid, and yet the warning is valid, so I'm silencing it
through push-disable-pop.
llvm-svn: 287391
