Respect alignment when loading up a coerced function argument
Code in CGCall.cpp that loads up function arguments that need to be
coerced to a different type may in some cases ignore the fact that
the source of the argument is not naturally aligned. This may cause
incorrect code to be generated. In some places in CreateCoercedLoad,
we already have setAlignment calls to address this, but I ran into one
where it was missing, causing wrong code generation on SystemZ.
However, in that location, we do not actually know what alignment of
the source location we can rely on; the callers do not pass anything
to this routine. This is already an issue in other places in
CreateCoercedLoad; and the same problem exists for CreateCoercedStore.
To avoid pessimising code, and to fix the FIXMEs already in place,
this patch also adds an alignment argument to the CreateCoerced*
routines and uses it instead of forcing an alignment of 1. The
callers are changed to pass in the best information they have.
This actually requires changes in a number of existing test cases
since we now get better alignment in many places.
Daniel Jasper [Fri, 10 Jul 2015 05:57:23 +0000 (05:57 +0000)]
Remove test that tests referring to the current working directory. You
cannot assume that the current working directory is writable in all test
environments. I don't know a better way to write this test of hand, lets
discuss. Possibly, a better option would be to put these together with
other test testing the driver directly.
CFI: Emit correct bit set information if RTTI is disabled under MS ABI.
We were previously creating bit set entries at virtual table offset
sizeof(void*) unconditionally under the Microsoft C++ ABI. This is incorrect
if RTTI data is disabled; in that case the "address point" is at offset
0. This change modifies bit set emission to take into account whether RTTI
data is being emitted.
Also make a start on a blacklisting scheme for records.
Diego Novillo [Thu, 9 Jul 2015 17:23:53 +0000 (17:23 +0000)]
Add GCC-compatible flags -fprofile-generate and -fprofile-use.
This patch adds support for specifying where the profile is emitted in a
way similar to GCC. These flags are used to specify directories instead
of filenames. When -fprofile-generate=DIR is used, the compiler will
generate code to write to <DIR>/default.profraw.
The patch also adds a couple of extensions: LLVM_PROFILE_FILE can still be
used to override the directory and file name to use and -fprofile-use
accepts both directories and filenames.
To simplify the set of flags used in the backend, all the flags get
canonicalized to -fprofile-instr-{generate,use} when passed to the
backend. The decision to use a default name for the profile is done
in the driver.
Add clang_free to libclang to free memory allocated in libclang.
One of the problems libclang tests has running under Windows is memory
allocated in libclang.dll but being freed in the test executable, possibly
by a different memory manager. This patch exposes a new export function,
clang_free(), used to free any allocated memory with the same libclang.dll
memory manager that allocated the memory.
Move the diagnostic back to codegen so that we can compile ATL on the
self-host bot. We don't actually end up emitting code for the __try, so
the diagnostic won't be hit.
Richard Smith [Wed, 8 Jul 2015 21:49:31 +0000 (21:49 +0000)]
[modules] Fix merging support for forward-declared enums with fixed underlying types. A visible declaration is enough to make the type complete, but not enough to make the definition visible.
Adrian Prantl [Wed, 8 Jul 2015 21:18:34 +0000 (21:18 +0000)]
Cleanup the doxygen comments in CGDebugInfo.cpp according to the coding
standards. Remove several hilariously out-of-date and redundant comments
and move the non-redundant ones into the header file.
Desugar doesn't necessarily initialize ShouldAKA, but as of r241542 it
may read it. Fix the misuse of the API and initialize this before
passing it in.
Petar Jovanovic [Wed, 8 Jul 2015 13:07:31 +0000 (13:07 +0000)]
[MIPS] Add support for direct-to-nacl in Clang
For Mips direct-to-nacl, the goal is to be close to le32 front-end and
use Mips32EL backend. This patch defines new NaClMips32ELTargetInfo and
modifies it slightly to be close to le32. It also adds necessary parts,
inline with ARM and X86.
[EH] Fix for clang bug 24005 - no cleanup for array of memcpy-able objects in struct (patch by Denis Zobnin)
The fix is to emit cleanup for arrays of memcpy-able objects in struct if an exception is thrown later during copy-construction.
Differential Revision: http://reviews.llvm.org/D10989
This reverts commit r239846 and r239879. They caused clang's
-fms-extensions behavior to incorrectly parse lambdas and includes a
testcase to ensure we don't regress again.
David Majnemer [Wed, 8 Jul 2015 05:07:05 +0000 (05:07 +0000)]
[CodeGen] Don't crash classifying a union of an AVX vector and an int
We forgot to run postMerge after decided that the union had to be
classified as MEMORY. This left us with Lo == MEMORY and Hi == SSEUp
which is an invalid combination.
Richard Smith [Wed, 8 Jul 2015 02:22:15 +0000 (02:22 +0000)]
[modules] When determining the visible module set during template
instantiation, use the set of modules visible from the template definition, not
from whichever declaration the specialization was instantiated from.
Adrian Prantl [Wed, 8 Jul 2015 01:39:38 +0000 (01:39 +0000)]
Change the expectation for test/Tooling/ms-asm-no-target.cpp since
clang-check now initializes the available targets to support
clang module containers.
Adrian Prantl [Tue, 7 Jul 2015 20:11:29 +0000 (20:11 +0000)]
Wrap clang modules and pch files in an object file container.
This patch adds ObjectFilePCHContainerOperations uses the LLVM backend
to put the contents of a PCH into a __clangast section inside a COFF, ELF,
or Mach-O object file container.
This is done to facilitate module debugging by makeing it possible to
store the debug info for the types defined by a module alongside the AST.
Appease MSVC 2013 a few more times, which was complaining about a syntax error with an unexpected close curly brace. For some reason, the range-based for loop requires curly braces to compile in MSVC.
Appease MSVC 2013, which was complaining about a syntax error with an unexpected close curly brace. For some reason, the range-based for loop requires curly braces to compile in MSVC.
Douglas Gregor [Tue, 7 Jul 2015 06:20:12 +0000 (06:20 +0000)]
The AST importer had a bug where it would enter into an infinite recursion
when importing type parameter lists. The reason is that type parameters
have their DeclContexts set to the interface that is parameterized with those
types, and the importer would follow that loop and blow the stack out.
I've changed the way this works so that the type parameters are only imported
after the interface that contains them has been registered via the Imported()
function.
Douglas Gregor [Tue, 7 Jul 2015 03:58:54 +0000 (03:58 +0000)]
Implement variance for Objective-C type parameters.
Introduce co- and contra-variance for Objective-C type parameters,
which allows us to express that (for example) an NSArray is covariant
in its type parameter. This means that NSArray<NSMutableString *> * is
a subtype of NSArray<NSString *> *, which is expected of the immutable
Foundation collections.
Type parameters can be annotated with __covariant or __contravariant
to make them co- or contra-variant, respectively. This feature can be
detected by __has_feature(objc_generics_variance). Implements
rdar://problem/20217490.
Douglas Gregor [Tue, 7 Jul 2015 03:58:42 +0000 (03:58 +0000)]
Implement the Objective-C __kindof type qualifier.
The __kindof type qualifier can be applied to Objective-C object
(pointer) types to indicate id-like behavior, which includes implicit
"downcasting" of __kindof types to subclasses and id-like message-send
behavior. __kindof types provide better type bounds for substitutions
into unspecified generic types, which preserves more type information.
Douglas Gregor [Tue, 7 Jul 2015 03:58:28 +0000 (03:58 +0000)]
Warn when an intended Objective-C specialization was actually a useless protocol qualification.
Warn in cases where one has provided redundant protocol qualification
that might be a typo for a specialization, e.g., NSArray<NSObject>,
which is pointless (NSArray declares that it conforms to NSObject) and
is likely to be a typo for NSArray<NSObject *>, i.e., an array of
NSObject pointers. This warning is very narrow, only applying when the
base type being qualified is parameterized, has the same number of
parameters as their are protocols listed, all of the names can also
refer to types (including Objective-C class types, of course), and at
least one of those types is an Objective-C class (making this a typo
for a missing '*'). The limitations are partly for performance reasons
(we don't want to do redundant name lookup unless we really need to),
and because we want the warning to apply in very limited cases to
limit false positives.
Douglas Gregor [Tue, 7 Jul 2015 03:58:22 +0000 (03:58 +0000)]
Warn when an Objective-C collection literal element is converted to an incompatible type.
Objective-C collection literals produce unspecialized
NSArray/NSDictionary objects that can then be implicitly converted to
specialized versions of these types. In such cases, check that the
elements in the collection are suitable for the specialized
collection. Part of rdar://problem/6294649.
Douglas Gregor [Tue, 7 Jul 2015 03:58:14 +0000 (03:58 +0000)]
C++ support for Objective-C lightweight generics.
Teach C++'s tentative parsing to handle specializations of Objective-C
class types (e.g., NSArray<NSString *>) as well as Objective-C
protocol qualifiers (id<NSCopying>) by extending type-annotation
tokens to handle this case. As part of this, remove Objective-C
protocol qualifiers from the declaration specifiers, which never
really made sense: instead, provide Sema entry points to make them
part of the type annotation token. Among other things, this properly
diagnoses bogus types such as "<NSCopying> id" which should have been
written as "id <NSCopying>".
Implements template instantiation support for, e.g., NSArray<T>*
in C++. Note that parameterized classes are not templates in the C++
sense, so that cannot (for example) be used as a template argument for
a template template parameter. Part of rdar://problem/6294649.
Douglas Gregor [Tue, 7 Jul 2015 03:58:01 +0000 (03:58 +0000)]
Improve the Objective-C common-type computation used by the ternary operator.
The Objective-C common-type computation had a few problems that
required a significant rework, including:
- Quadradic behavior when finding the common base type; now it's
linear.
- Keeping around type arguments when computing the common type
between a specialized and an unspecialized type
- Introducing redundant protocol qualifiers.
Part of rdar://problem/6294649. Also fixes rdar://problem/19572837 by
addressing a longstanding bug in
ASTContext::CollectInheritedProtocols().
Douglas Gregor [Tue, 7 Jul 2015 03:57:53 +0000 (03:57 +0000)]
Substitute type arguments into uses of Objective-C interface members.
When messaging a method that was defined in an Objective-C class (or
category or extension thereof) that has type parameters, substitute
the type arguments for those type parameters. Similarly, substitute
into property accesses, instance variables, and other references.
This includes general infrastructure for substituting the type
arguments associated with an ObjCObject(Pointer)Type into a type
referenced within a particular context, handling all of the
substitutions required to deal with (e.g.) inheritance involving
parameterized classes. In cases where no type arguments are available
(e.g., because we're messaging via some unspecialized type, id, etc.),
we substitute in the type bounds for the type parameters instead.
When substituting for the type parameters given an unspecialized
context (i.e., no specific type arguments were given), substituting
the type bounds unconditionally produces type signatures that are too
strong compared to the pre-generics signatures. Instead, use the
following rule:
- In covariant positions, such as method return types, replace type
parameters with “id” or “Class” (the latter only when the type
parameter bound is “Class” or qualified class, e.g,
“Class<NSCopying>”)
- In other positions (e.g., parameter types), replace type
parameters with their type bounds.
- When a specialized Objective-C object or object pointer type
contains a type parameter in its type arguments (e.g.,
NSArray<T>*, but not NSArray<NSString *> *), replace the entire
object/object pointer type with its unspecialized version (e.g.,
NSArray *).
Douglas Gregor [Tue, 7 Jul 2015 03:57:35 +0000 (03:57 +0000)]
Handle Objective-C type arguments.
Objective-C type arguments can be provided in angle brackets following
an Objective-C interface type. Syntactically, this is the same
position as one would provide protocol qualifiers (e.g.,
id<NSCopying>), so parse both together and let Sema sort out the
ambiguous cases. This applies both when parsing types and when parsing
the superclass of an Objective-C class, which can now be a specialized
type (e.g., NSMutableArray<T> inherits from NSArray<T>).
Check Objective-C type arguments against the type parameters of the
corresponding class. Verify the length of the type argument list and
that each type argument satisfies the corresponding bound.
Specializations of parameterized Objective-C classes are represented
in the type system as distinct types. Both specialized types (e.g.,
NSArray<NSString *> *) and unspecialized types (NSArray *) are
represented, separately.
Douglas Gregor [Tue, 7 Jul 2015 03:57:15 +0000 (03:57 +0000)]
Parsing, semantic analysis, and AST for Objective-C type parameters.
Produce type parameter declarations for Objective-C type parameters,
and attach lists of type parameters to Objective-C classes,
categories, forward declarations, and extensions as
appropriate. Perform semantic analysis of type bounds for type
parameters, both in isolation and across classes/categories/extensions
to ensure consistency.
Also handle (de-)serialization of Objective-C type parameter lists,
along with sundry other things one must do to add a new declaration to
Clang.
Note that Objective-C type parameters are typedef name declarations,
like typedefs and C++11 type aliases, in support of type erasure.
Adrian Prantl [Tue, 7 Jul 2015 00:49:35 +0000 (00:49 +0000)]
Debug info: Emit distinct __block_literal_generic types for blocks with
different function signatures. (Previously clang would emit all block
pointer types with the type of the first block pointer in the compile
unit.)
This reverts commit r241244, but restricts SEH support to Win64.
This way, Chromium builds will still fall back on TUs with SEH, and
Clang developers can work on this incrementally upstream while patching
this small predicate locally. It'll also make it easier to review small
fixes.