Eli Friedman [Tue, 14 Jun 2011 01:37:52 +0000 (01:37 +0000)]
The LLVM IR representation of byval arguments has a rather strange property: if the alignment of an argument to a call is less than the specified byval alignment for that argument, there is no way to specify the alignment of the implied copy. Therefore, we must ensure that the alignment of the argument is at least the byval alignment. To do this, we have to mess with the alignment of relevant alloca's in some cases, and insert a copy that conceptually shouldn't be necessary in some cases.
This patch tries relatively hard to avoid creating an extra copy if it can be avoided (see test3 in the included testcase), but it is not possible to avoid in some cases (like test2 in the included testcase).
Richard Smith [Sun, 12 Jun 2011 11:43:46 +0000 (11:43 +0000)]
Don't assert on initialized typedef declarations in classes:
struct {
typedef int A = 0;
};
According to the C++11 standard, this is not ill-formed, but does not have any ascribed meaning. We can't reasonably accept it, so treat it as ill-formed.
Also switch C++ from an incorrect 'fields can only be initialized in constructors' diagnostic for this case to C's 'illegal initializer (only variables can be initialized)'
Douglas Gregor [Sat, 11 Jun 2011 01:09:30 +0000 (01:09 +0000)]
Implement Objective-C Related Result Type semantics.
Related result types apply Cocoa conventions to the type of message
sends and property accesses to Objective-C methods that are known to
always return objects whose type is the same as the type of the
receiving class (or a subclass thereof), such as +alloc and
-init. This tightens up static type safety for Objective-C, so that we
now diagnose mistakes like this:
t.m:4:10: warning: incompatible pointer types initializing 'NSSet *'
with an
expression of type 'NSArray *' [-Wincompatible-pointer-types]
NSSet *array = [[NSArray alloc] init];
^ ~~~~~~~~~~~~~~~~~~~~~~
/System/Library/Frameworks/Foundation.framework/Headers/NSObject.h:72:1:
note:
instance method 'init' is assumed to return an instance of its
receiver
type ('NSArray *')
- (id)init;
^
It also means that we get decent type inference when writing code in
Objective-C++0x:
auto array = [[NSMutableArray alloc] initWithObjects:@"one", @"two",nil];
// ^ now infers NSMutableArray* rather than id
Eli Friedman [Fri, 10 Jun 2011 21:53:06 +0000 (21:53 +0000)]
PR10120: Make CodeGenModule::getVTableLinkage use NamedDecl::getLinkage to determine whether the vtable should be externally visible, instead of a rough approximation of it which messes up with templates.
While I'm here, zap the other user of isInAnonymousNamespace outside of Decl.cpp.
Sean Hunt [Fri, 10 Jun 2011 09:24:41 +0000 (09:24 +0000)]
Implement caching of copy assignment operator lookup.
I believe, upon, careful review, that this code causes us to incorrectly
handle exception specifications of copy assignment operators in C++03
mode. However, we currently do not seem to properly implement the subtle
distinction between copying of members and bases made by implicit copy
constructors and assignment operators in C++03 - namely that they are
limited in their overload selection - in all cases. As such, I feel that
committing this code is correct pending a careful review of our
implementation of these semantics.
Jordy Rose [Fri, 10 Jun 2011 08:49:37 +0000 (08:49 +0000)]
[analyzer] PR8962 again. Ban ParenExprs (and friends) from block-level expressions (by calling IgnoreParens before adding expressions to blocks). Undo 132769 (LiveVariables' local IgnoreParens), since it's no longer necessary.
Also, have Environment stop looking through NoOp casts; it didn't match the behavior of LiveVariables. And once that's gone, the whole cast block of that switch is unnecessary.
Richard Trieu [Fri, 10 Jun 2011 03:11:26 +0000 (03:11 +0000)]
Made changes to how 'struct'/'class' mismatches are handled in -Wmismatched-tags.
- Removed fix-it hints from template instaniations since changes to the
templates are rarely helpful.
- Changed the caret in template instaniations from the class/struct name to the
class/struct keyword, matching the other warnings.
- Do not offer fix-it hints when multiple declarations disagree. Warnings are
still given.
- Once a definition is found, offer a fix-it hint to all previous declarations
with wrong tag.
- Declarations that disagree with a previous definition will get a fix-it hint
to change the declaration.
Fix a regression in the two-phase lookup diagnostics that switching the
namespace set algorithm (re-)introduced. We may not have seen the 'std'
namespace, but we should still suggested associated namespaces. Easy
fix, but a bit annoying to test.
Nick Lewycky [Tue, 7 Jun 2011 06:07:12 +0000 (06:07 +0000)]
The macros defined by the language standard are still available even when the
-undef flag is passed in. Also __ASSEMBLER__ with -x assembler-with-cpp. (Don't
ask.)
Douglas Gregor [Tue, 7 Jun 2011 02:21:18 +0000 (02:21 +0000)]
Downgrade the warning about the use of typedefs for class template
specializations within an explicit instantiation to default to off
(enabled by -pedantic). Nobody else seem to implement C++
[temp.explicit]p3. Fixes PR10093.
Douglas Gregor [Mon, 6 Jun 2011 15:22:55 +0000 (15:22 +0000)]
Diagnose the condition in C++ [temp.expl.spec]p16 that prohibits
specializing a member of an unspecialized template, and recover from
such errors without crashing. Fixes PR10024 / <rdar://problem/9509761>.
Richard Smith [Sun, 5 Jun 2011 22:42:48 +0000 (22:42 +0000)]
Fix PR10053: Improve diagnostics and error recovery for code which some compilers incorrectly accept due to a lack of proper support for two-phase name lookup.
Sebastian Redl [Sun, 5 Jun 2011 13:59:11 +0000 (13:59 +0000)]
Drop most of InitializationSequence::SequenceKind's values. They didn't really contain any information that the step array didn't contain too. This makes debugging dumps a bit less informative, but probably not significantly so. The advantage is that the redundancy is gone, so the code is easier to understand.
ReferenceBinding is still there, because it is used in some unclear code.
Sebastian Redl [Sun, 5 Jun 2011 12:23:28 +0000 (12:23 +0000)]
Remove all references to InitializationSequence::FailedSequence from outside SemaInit.cpp. Replace them with the boolean conversion or the new Failed() function. This is a first step towards removing InitializationSequence::SequenceKind. No functionality change.
Douglas Gregor [Sun, 5 Jun 2011 06:15:20 +0000 (06:15 +0000)]
Identity and non-identity standard conversion sequences can be
compared even when one is a reference binding and the other is not
(<rdar://problem/9173984>), but the definition of an identity sequence
does not involve lvalue-to-rvalue adjustments (PR9507). Fix both
inter-related issues.
Douglas Gregor [Sun, 5 Jun 2011 05:04:23 +0000 (05:04 +0000)]
When inferring the result type of a block based on a return statement
with a type-dependent expression, infer the placeholder type
'Context.DependentTy' to indicate that this is just a
placeholder. Fixes PR9982 / <rdar://problem/9486685>.
Nick Lewycky [Sat, 4 Jun 2011 05:19:42 +0000 (05:19 +0000)]
Identifiers with _CapitalLetter are reserved, so don't use them. Prefer the
common C++ pattern of using the same name for the constructor argument as you
do for the member. Noticed by inspection.
Jordy Rose [Sat, 4 Jun 2011 01:47:27 +0000 (01:47 +0000)]
[analyzer] Don't crash when copying an unknown number of bytes with memcpy(). Also handle all memcpy-family return values in evalCopyCommon(), rather than having some outside and some inside.
Tanya Lattner [Sat, 4 Jun 2011 00:47:47 +0000 (00:47 +0000)]
Add support for builtin astype:
__builtin_astype(): Used to reinterpreted as another data type of the same size using for both scalar and vector data types.
Added test case.
Devang Patel [Sat, 4 Jun 2011 00:38:02 +0000 (00:38 +0000)]
Rework r132576.
Emit debug info only if there is an insertion point. The debug info should not force an insertion point. Codegen may later on decide to not emit code for some reason, see extensive comment in CodeGenFunction::EmitStmt(), and debug info should not get in the way.
Improvements to abbreviations for PCH which add support for EnumDecl, ObjCIvarDecl, TypedefDecl, VarDecl and FieldDecl and improve support for ParmVarDecl.
projects / clang / log