[preprocessor] Fix assertion hit when 'SingleFileParseMode' option is enabled and #if with an undefined identifier and without #else
'HandleEndifDirective' asserts that 'WasSkipping' is false, so switch to using 'FoundNonSkip' as the hint for 'SingleFileParseMode' to keep going with parsing.
Rui Ueyama [Wed, 21 Jun 2017 16:50:38 +0000 (16:50 +0000)]
Use -NOT prefix instead of adding `not` to FileCheck.
If we want to make sure that a particular string is not in an output,
the regular way of doing it is to add `-NOT` prefix instead of checking
if FileCheck resulted in an error.
Erich Keane [Wed, 21 Jun 2017 16:37:22 +0000 (16:37 +0000)]
Correct VectorCall x86 (32 bit) behavior for SSE Register Assignment
In running some internal vectorcall tests in 32 bit mode, we discovered that the
behavior I'd previously implemented for x64 (and applied to x32) regarding the
assignment of SSE registers was incorrect. See spec here:
https://msdn.microsoft.com/en-us/library/dn375768.aspx
My previous implementation applied register argument position from the x64
version to both. This isn't correct for x86, so this removes and refactors that
section. Additionally, it corrects the integer/int-pointer assignments. Unlike
x64, x86 permits integers to be assigned independent of position.
Finally, the code for 32 bit was cleaned up a little to clarify the intent,
as well as given a descriptive comment.
Francois Ferrand [Wed, 21 Jun 2017 13:56:02 +0000 (13:56 +0000)]
clang-format: introduce InlineOnly short function style
Summary:
This is the same as Inline, except it does not imply all empty
functions are merged: with this style, empty functions are merged only
if they also match the 'inline' criteria (i.e. defined in a class).
This is helpful to avoid inlining functions in implementations files.
Serge Pavlov [Wed, 21 Jun 2017 12:46:57 +0000 (12:46 +0000)]
Function with unparsed body is a definition
While a function body is being parsed, the function declaration is not considered
as a definition because it does not have a body yet. In some cases it leads to
incorrect interpretation, the case is presented in
https://bugs.llvm.org/show_bug.cgi?id=14785:
```
template<typename T> struct Somewhat {
void internal() const {}
friend void operator+(int const &, Somewhat<T> const &) {}
};
void operator+(int const &, Somewhat<char> const &x) { x.internal(); }
```
When statement `x.internal()` in the body of global `operator+` is parsed, the type
of `x` must be completed, so the instantiation of `Somewhat<char>` is started. It
instantiates the declaration of `operator+` defined inline, and makes a check for
redefinition. The check does not detect another definition because the declaration
of `operator+` is still not defining as does not have a body yet.
To solves this problem the function `isThisDeclarationADefinition` considers
a function declaration as a definition if it has flag `WillHaveBody` set.
Alex Lorenz [Wed, 21 Jun 2017 10:27:24 +0000 (10:27 +0000)]
Revert r305678: [driver][macOS] Pick the system version for the
deployment target if the SDK is newer than the system
This commit also reverts follow-up commits r305680 and r305685 that have
buildbot fixes.
The change in r305678 wasn't correct because it relied on
`llvm::sys::getProcessTriple`, which uses a pre-configured OS version. We should
lookup the actual macOS version of the system on which the compiler is running.
Support MS builtins using 'long' on LP64 platforms
This allows for -fms-extensions to work the same on LP64. For example,
_BitScanReverse is expected to be 32-bit, matching Windows/LLP64, even
though long is 64-bit on x86_64 Darwin or Linux (LP64).
Implement this by adding a new character code 'N', which is 'int' if
the target is LP64 and the same 'L' otherwise
[clang] Fix format specifiers fixits for nested macros
ExpansionLoc was previously calculated incorrectly in the case of
nested macros expansions. In this diff we build the stack of expansions
where the last one is the actual expansion which should be used
for grouping together the edits.
The definition of MacroArgUse is adjusted accordingly.
Vassil Vassilev [Tue, 20 Jun 2017 14:59:57 +0000 (14:59 +0000)]
D31187: Fix removal of out-of-line definitions.
Consider:
struct MyClass {
void f() {}
}
MyClass::f(){} // expected error redefinition of f. #1
Some clients (eg. cling) need to call removeDecl for the redefined (#1) decl.
This patch enables us to remove the lookup entry is registered in the semantic
decl context and not in the primary decl context of the lexical decl context
where we currently are trying to remove it from.
It is not trivial to test this piece and writing a full-blown unit test seems
too much.
[preprocessor] When preprocessor option 'SingleFileParseMode' is enabled, parse all directive blocks if the condition uses undefined macros
This is useful for being able to parse the preprocessor directive blocks even if the header, that defined the macro that is checked, hasn't been included.
Manoj Gupta [Mon, 19 Jun 2017 18:45:03 +0000 (18:45 +0000)]
[Clang] Handle interaction of -pg and no_instrument_function attribute.
Summary:
Disable generation of counting-function attribute if no_instrument_function
attribute is present in function.
Interaction between -pg and no_instrument_function is the desired behavior
and matches gcc as well.
This is required for fixing a crash in Linux kernel when function tracing
is enabled.
Yaxun Liu [Mon, 19 Jun 2017 17:03:41 +0000 (17:03 +0000)]
CodeGen: Cast temporary variable to proper address space
In C++ all variables are in default address space. Previously change has been
made to cast automatic variables to default address space. However that is
not sufficient since all temporary variables need to be casted to default
address space.
This patch casts all temporary variables to default address space except those
for passing indirect arguments since they are only used for load/store.
This patch only affects target having non-zero alloca address space.
Summary:
This fixes the missing space before the designated initializer when `Cpp11BracedListStyle=false` :
const struct A a = { .a = 1, .b = 2 };
^
Also, wrapping between opening brace and designated array initializers used to have an excessive penalty (like breaking between an expression and the subscript operator), leading to unexpected wrapping:
Christof Douma [Mon, 19 Jun 2017 12:05:58 +0000 (12:05 +0000)]
[NFC] Refactor DiagnosticRenderer to use FullSourceLoc
Move the DiagnosticRenderer and its dependents to using FullSourceLocs
instead of a SourceLocation and SourceManager pointer. The changeset is
rather large but entirely mechanical.
This is step one to allow DiagnosticRenderer to take either
llvm::SMLocs or clang::SourceLocations.
Alex Lorenz [Mon, 19 Jun 2017 10:57:27 +0000 (10:57 +0000)]
[driver][macOS] Pick the system version for the deployment target
if the SDK is newer than the system
This commit improves the driver by making sure that it picks the system version
for the deployment target when the version of the macOS SDK is newer than the
system version.
Daniel Jasper [Mon, 19 Jun 2017 07:40:49 +0000 (07:40 +0000)]
clang-format: Handle "if constexpr".
c++1z adds the following constructions to the language:
if constexpr (cond)
statement1;
else if constexpr (cond)
statement2;
else if constexpr (cond)
statement3;
else
statement4;
A first version of this was proposed in reviews.llvm.org/D26953 by
Francis Visoiu Mistrih, but never commited. This patch additionally
fixes the behavior when allowing short if statements on a single line
and was authored by Jacob Bandes-Storch. Thank you to both authors.
Daniel Jasper [Mon, 19 Jun 2017 07:30:04 +0000 (07:30 +0000)]
clang-format: Add capability to format the diff on save in vim.
With this patch, one can configure a BufWrite hook that will make the
clang-format integration compute a diff of the current buffer with the file
that's on disk and format all changed lines. This should create a
zero-overhead auto-format solution that doesn't require the file to
already be clang-format clean to avoid spurious diffs.
CodeGen: make the type match the comment for a libcall
Fix the type for a (runtime) library call to match both the comment and
the runtime implementation. As it happens, the type being used matched,
this just makes it more precise.
Lang Hames [Sat, 17 Jun 2017 00:12:38 +0000 (00:12 +0000)]
Call setMustBuildLookupTable on TagDecls in ExternalASTMerger
Summary:
setMustBuildLookupTable should be called on imported TagDecls otherwise we may fail
to import their member decls (if they have any).
Not calling the setMustBuildLookupTable method results in a failure in the attached test
case when lookup for the 'x' member fails on struct S, which hasn't had its decls imported
elsewhere. (By contrast the member-in-struct testcase hasn't run into this issue
because the import of its decls is triggered when the struct instance is defined, and the
member access follows this).
Petr Hosek [Fri, 16 Jun 2017 22:40:18 +0000 (22:40 +0000)]
[Driver] Do a PATH lookup if needed when using -no-canonical-prefixes
When -no-canonical-prefixes option is used and argv0 contains only
a program name, we need to do a PATH lookup to get an executable path,
otherwise the return value won't be a valid path and any subsequent
uses of it (e.g. when invoking -cc1) will fail with an error.
Alex Lorenz [Fri, 16 Jun 2017 20:13:39 +0000 (20:13 +0000)]
[PR33394] Avoid lexing editor placeholders when Clang is used only
for preprocessing
r300667 added support for editor placeholder to Clang. That commit didn’t take
into account that users who use Clang for preprocessing only (-E) will get the
"editor placeholder in source file" error when preprocessing their source
(PR33394). This commit ensures that Clang doesn't lex editor placeholders when
running a preprocessor only action.
[XRay][clang] Support capturing the implicit `this` argument to C++ class member functions
Summary:
Before this change, we couldn't capture the `this` pointer that's
implicitly the first argument of class member functions. There are some
interesting things we can do with capturing even just this single
argument for zero-argument member functions.
Richard Trieu [Fri, 16 Jun 2017 02:44:29 +0000 (02:44 +0000)]
[ODRHash] Hash VarDecl members.
These VarDecl's are static data members of classes. Since the initializers are
also hashed, this also provides checking for default arguments to methods.
Alex Lorenz [Thu, 15 Jun 2017 21:40:54 +0000 (21:40 +0000)]
[Completion] Code complete the members for a dependent type after a '::'
This commit is a follow up to r302797 which added support for dependent
completions after the '.' and '->' operators. This commit adds support for
dependent completions after the '::' operator.
Alex Lorenz [Thu, 15 Jun 2017 20:50:43 +0000 (20:50 +0000)]
[index] Index static_assert declarations
static_assert declarations have to be visited while indexing so that we can
gather the references to declarations that are present in their assert
expression.
Apply summary-based dead stripping to regular LTO modules with summaries.
If a regular LTO module has a summary index, then instead of linking
it into the combined regular LTO module right away, add it to the
combined summary index and associate it with a special module that
represents the combined regular LTO module.
Any such modules are linked during LTO::run(), at which time we use
the results of summary-based dead stripping to control whether to
link prevailing symbols.
Dominic Chen [Thu, 15 Jun 2017 17:05:07 +0000 (17:05 +0000)]
[analyzer]: Improve test handling with multiple constraint managers
Summary: Modify the test infrastructure to properly handle tests that require z3, and merge together the output of all tests on success. This is required for D28954.
Daniel Jasper [Thu, 15 Jun 2017 09:17:12 +0000 (09:17 +0000)]
Revert "Define _GNU_SOURCE for rtems c++"
This reverts commit r305399.
This breaks a build in libcxx:
libcxx/src/system_error.cpp:90:16: error: assigning to 'int' from incompatible type 'char *'
if ((ret = ::strerror_r(ev, buffer, strerror_buff_size)) != 0) {
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 error generated.
Which makes sense according to:
https://linux.die.net/man/3/strerror_r
Erich Keane [Wed, 14 Jun 2017 23:09:01 +0000 (23:09 +0000)]
[Preprocessor]Correct Macro-Arg allocation of StringifiedArguments,
correct getNumArguments
StringifiedArguments is allocated (resized) based on the size the
getNumArguments function. However, this function ACTUALLY currently
returns the amount of total UnexpArgTokens which is minimum the same as
the new implementation of getNumMacroArguments, since empty/omitted arguments
result in 1 UnexpArgToken, and included ones at minimum include 2
(1 for the arg itself, 1 for eof).
This patch renames the otherwise unused getNumArguments to be more clear
that it is the number of arguments that the Macro expects, and thus the maximum
number that can be stringified. This patch also replaces the explicit memset
(which results in value instantiation of the new tokens, PLUS clearing the
memory) with brace initialization.
Serge Pavlov [Wed, 14 Jun 2017 10:07:02 +0000 (10:07 +0000)]
Function with unparsed body is a definition
While a function body is being parsed, the function declaration is not considered
as a definition because it does not have a body yet. In some cases it leads to
incorrect interpretation, the case is presented in
https://bugs.llvm.org/show_bug.cgi?id=14785:
```
template<typename T> struct Somewhat {
void internal() const {}
friend void operator+(int const &, Somewhat<T> const &) {}
};
void operator+(int const &, Somewhat<char> const &x) { x.internal(); }
```
When statement `x.internal()` in the body of global `operator+` is parsed, the type
of `x` must be completed, so the instantiation of `Somewhat<char>` is started. It
instantiates the declaration of `operator+` defined inline, and makes a check for
redefinition. The check does not detect another definition because the declaration
of `operator+` is still not defining as does not have a body yet.
To solves this problem the function `isThisDeclarationADefinition` considers
a function declaration as a definition if it has flag `WillHaveBody` set.
Eric Fiselier [Wed, 14 Jun 2017 03:24:55 +0000 (03:24 +0000)]
[coroutines] Fix co_await for range statement
Summary:
Currently we build the co_await expressions on the wrong implicit statements of the implicit ranged for; Specifically we build the co_await expression wrapping the range declaration, but it should wrap the begin expression.
projects / clang / log