Remove TransferFuncs from ExprEngine and AnalysisConsumer.
Demote RetainReleaseChecker to a regular checker, and give it the name osx.cocoa.RetainCount (class name change coming shortly). Update tests accordingly.
llvm-svn: 138998
include guards don't show up as macro definitions in every translation
unit that imports a module. Macro definitions can, however, be
exported with the intentionally-ugly #__export_macro__
directive. Implement this feature by not even bothering to serialize
non-exported macros to a module, because clients of that module need
not (should not) know that these macros even exist.
llvm-svn: 138943
I had to force -fno-delayed-template-parsing on some Index tests because delayed template parsing will change the output of some tests.
llvm-svn: 138942
semantic analysis when taking the address of an xvalue. Instead, just
build the unary operator directly, since it's safe to do so (from the
IRgen and AST perspectives) for any glvalue. Fixes PR10822.
llvm-svn: 138935
The initial incentive was to fix a crash when PCH chaining categories
to an interface, but the fix was done in the "modules way" that I hear
is popular with the kids these days.
Each module stores the local chain of categories and we combine them
when the interface is loaded. We also warn if non-dependent modules
introduce duplicate named categories.
llvm-svn: 138926
synthesis. This new feature is currently placed under
-fobjc-default-synthesize-properties option
and is off by default pending further testing.
It will become the default feature soon.
// rdar://8843851
llvm-svn: 138913
existing practice with Python extension modules. Not that Python
extension modules should be using a double-underscored identifier
anyway, but...
llvm-svn: 138870
and does the Sema on their body after the entire
class/category @implementation is seen. This change allows messaging
of forward private methods, as well as, access to
synthesized ivars of properties with foward synthesize
declarations; among others. In effect, this patch removes
several restrictions placed on objective-c due to in-place
semantics processing of methods.
This is part of // rdar://8843851.
llvm-svn: 138865
of the function in question when applicable (that is, not for blocks).
Patch by Joerg Sonnenberger with some stylistic tweaks by me.
When discussing this weth Joerg, streaming the decl directly into the
diagnostic didn't work because we have a pointer-to-const, and the
overload doesn't accept such. In order to make my style tweaks to the
patch, I first changed the overload to accept a pointer-to-const, and
then changed the diagnostic printing layer to also use
a pointer-to-const, cleaning up a gross line of code along the way.
llvm-svn: 138854
collision between C99 hexfloats and C++0x user-defined literals by
giving C99 hexfloats precedence. Also, warning about user-defined
literals that conflict with hexfloats and those that have names that
are reserved by the implementation. Fixes <rdar://problem/9940194>.
llvm-svn: 138839
sure that all of the CXXConversionDecls go into the same
bucket. Otherwise, name lookup might not find them all. Fixes
<rdar://problem/10041960>.
llvm-svn: 138824
This makes the code duplication of implicit special member handling even worse,
but the cleanup will have to come later. For now, this works.
Follow-up with tests for explicit defaulting and enabling the __has_feature
flag to come.
llvm-svn: 138821
- wrong alignment for double (it was 4, but 8 is desired),
- added checks for _REENTRANT define,
- fixed the issue that defines were not tested (because the check for inside #ifdef).
llvm-svn: 138775
qualification of a type doesn't affect whether a conversion is a narrowing
conversion.
This doesn't work in template cases because SubstTemplateTypeParmType gets in
the way.
llvm-svn: 138735
to varargs functions in unevaluated contexts. AFAICT, there is no
standards justification for this, but it matches what other compilers do
and therefore preserves compatibility with certain template metaprogramming
idioms.
Should fix self-host.
llvm-svn: 138715
, such as list of forward @class decls, in a DeclGroup
node. Deal with its consequence throught clang. This
is in preparation for more Sema work ahead. // rdar://8843851.
Feel free to reverse if it breaks something important
and I am unavailable.
llvm-svn: 138709
__import__ within the preprocessor, since the prior one foolishly
assumed that Preprocessor::Lex() was re-entrant. We now handle
__import__ at the top level (only), after macro expansion. This should
fix the buildbot failures.
llvm-svn: 138704
loads the named module. The syntax itself is intentionally hideous and
will be replaced at some later point with something more
palatable. For now, we're focusing on the semantics:
- Module imports are handled first by the preprocessor (to get macro
definitions) and then the same tokens are also handled by the parser
(to get declarations). If both happen (as in normal compilation),
the second one is redundant, because we currently have no way to
hide macros or declarations when loading a module. Chris gets credit
for this mad-but-workable scheme.
- The Preprocessor now holds on to a reference to a module loader,
which is responsible for loading named modules. CompilerInstance is
the only important module loader: it now knows how to create and
wire up an AST reader on demand to actually perform the module load.
- We search for modules in the include path, using the module name
with the suffix ".pcm" (precompiled module) for the file name. This
is a temporary hack; we hope to improve the situation in the
future.
llvm-svn: 138679
These tests have been failing since the beginning of buildbot time:
crash-recovery-code-complete.c
crash-recovery-reparse.c
crash-recovery.c
llvm-svn: 138672
really shouldn't be optional. Fix the remaining place where a
temporary was being passed as potentially-aliased memory.
Fixes PR10756.
llvm-svn: 138627
Also convert stack-addr-ps.cpp to use the analyzer instead of just Sema, now
that it doesn't crash, and extract the stack-block test into another file since
it errors, and that prevents the analyzer from running.
llvm-svn: 138613
Much to everyone's surprise, the default constructor for TypeResult produces
an instance with Invalid == false. This seems like a decision we may want to
revisit.
llvm-svn: 138601
emit call results into potentially aliased slots. This allows us
to properly mark indirect return slots as noalias, at the cost
of requiring an extra memcpy when assigning an aggregate call
result into a l-value. It also brings us into compliance with
the x86-64 ABI.
llvm-svn: 138599
from the given source. -emit-module behaves similarly to -emit-pch,
except that Sema is somewhat more strict about the contents of
-emit-module. In the future, there are likely to be more interesting
differences.
llvm-svn: 138595
of a type specifier.
e.g. for:
typedef struct _MyS {
int foo;
} MyS;
pointing at field 'foo' would give a cursor for the typedef declaration 'MyS'
instead of the field.
llvm-svn: 138593
redeclarations of a particular entity would occur in source
order. Friend declarations that occur within class templates (or
member classes thereof) do not follow this, nor would modules. Big
thanks to Erik Verbruggen for reducing this problem from the Very
Large Qt preamble testcase he found.
llvm-svn: 138557
given selector, rather than walking the chain backwards. Teach its
visitor how to merge multiple result sets into a single result set,
combining the results of selector lookup in several different modules
into a single result set.
llvm-svn: 138556
which supports both pre-order and post-order traversal via a visitor
mechanism. Use this depth-first search with a post-order traversal to
give predictable ordering semantics when walking all of the lexical
declarations in the translation unit.
Eventually, module imports will occur in the source code rather than
at the beginning, and we'll have to revisit this walk.
llvm-svn: 138490
module DAG-based lookup scheme. This required some reshuffling, so
that each module stores its own mapping from DeclContexts to their
lexical and visible sets for those DeclContexts (rather than one big
"chain").
Overall, this allows simple qualified name lookup into the translation
unit to gather results from multiple modules, with the lookup results
in module B shadowing the lookup results in module A when B imports A.
Walking all of the lexical declarations in a module DAG is still a
mess; we'll end up walking the loaded module list backwards, which
works fine for chained PCH but doesn't make sense in a DAG. I'll
tackle this issue as a separate commit.
llvm-svn: 138463
incorrectly in the CFG, and also the static analyzer. This patch regresses the analyzer a bit, but
that needs to be followed up with a better solution.
Fixes <rdar://problem/10008112>.
llvm-svn: 138372
system flags an error when unlocking a lock which was not held, locking
the same lock twice, having a different lockset on each iteration of a
loop, or going out of scope while still holding a lock. In order to
successfully use the lockset, this patch also makes sure that attribute
arguments are attached correctly for later parsing.
This patch was also worked on by DeLesley Hutchins.
Note: This patch has been reviewed by Chandler Carruth and Jeffrey
Yasskin. Feel free to provide post-commit review comments for a
subsequent patch.
llvm-svn: 138350
output on darwin so is hard coded there.
As a note this will need a little bit of refactoring in the class
hierarchy to separate it out for different verifiers based on input type.
Fixes rdar://8256258.
llvm-svn: 138343
For the test case added to function-redecl.cpp, we were previously complaining
about a mismatch in the parameter types, since the definition used the
typedef'd type.
llvm-svn: 138318
const int &x = x;
This crashed by inifinetly recursing within the lvalue evaluation
routine. I've added a (somewhat) braindead way of preventing this
recursion. If folks have better suggestions for how to avoid it I'm all
ears.
That said, we have some work to do. This doesn't trigger a single
warning for uninitialized, self-initialized or otherwise completely
wrong code. In some senses, the crash was almost better.
llvm-svn: 138239
like to avoid within the Clang test suite, it doesn't verify the output
at all so it only servers as a test that Clang doesn't crash, and
finally all it does is declare a function that returns a vector and call
it. Probably the biggest thing being tested here is Clang's parsing of
the vector template, and we have lots of good template parsing tests. We
don't need another in codegen.
llvm-svn: 138197
different modules) more robust. It already handled (simple) merges of
the set of declarations attached to that identifier, so add a test
case that shows us getting two different declarations for the same
identifier (one struct, one function) from different modules, and are
able to use both of them.
llvm-svn: 138189
modules (those that no other module depends on) and performs a search
over all of the modules, visiting a new module only when all of the
modules that depend on it have already been visited. The visitor can
abort the search for all modules that a module depends on, which
allows us to minimize the number of lookups necessary when performing
a search.
Switch identifier lookup from a linear walk over the set of modules to
this module visitation operation. The behavior is the same for simple
PCH and chained PCH, but provides the proper search order for
modules. Verified with printf debugging, since we don't have enough in
place to actually test this.
llvm-svn: 138187
to modernity. Instead of passing down individual
context objects from parser to sema, establish decl
context in parser and have sema access current context
as needed. I still need to take of Doug's comment for
minor cleanups.
llvm-svn: 138040
has already been loaded before allocating a new Module structure. If
the module has already been loaded (uniquing based on file name), then
just return the existing module rather than trying to load it again.
This allows us to load a DAG of modules. Introduce a simple test case
that forms a diamond-shaped module graph, and illustrates that a
source file importing the bottom of the diamond can see declarations
in all four of the modules that make up the diamond.
Note that this version moves the file-opening logic into the module
manager, rather than splitting it between the module manager and the
AST reader. More importantly, it properly handles the
weird-but-possibly-useful case of loading an AST file from "-".
llvm-svn: 138030
Teach ModuleManager::addModule() to check whether a particular module
has already been loaded before allocating a new Module structure. If
the module has already been loaded (uniquing based on file name), then
just return the existing module rather than trying to load it again.
This allows us to load a DAG of modules. Introduce a simple test case
that forms a diamond-shaped module graph, and illustrates that a
source file importing the bottom of the diamond can see declarations
in all four of the modules that make up the diamond.
llvm-svn: 137971
implicitly instantiable, even if we don't see a body on the friend
function declaration. The body may simply have not yet been attached.
This fixes PR10666.
There may be an alternate, preferred implementation strategy, see my
FIXME. Review would definitely be appreciated Doug. =D
llvm-svn: 137934
has already been loaded before allocating a new Module structure. If
the module has already been loaded (uniquing based on file name), then
just return the existing module rather than trying to load it again.
This allows us to load a DAG of modules. Introduce a simple test case
that forms a diamond-shaped module graph, and illustrates that a
source file importing the bottom of the diamond can see declarations
in all four of the modules that make up the diamond.
llvm-svn: 137925
Currently this includes -pedantic warnings as well; we'll need to consider whether these should
be included.
This works as expected with -Werror.
Test cases were added to Sema/warn-unused-parameters.c, but they should probably be broken off into
their own test file.
llvm-svn: 137910
automatically invoking llvm-gcc's cc1plus, which doesn't support all options
supported by Clang. Therefore, filter out unsupported options.
rdar://9964354
llvm-svn: 137842
even when overloaded and user-defined. These operators are both more
valuable to warn on (due to likely typos) and extremely unlikely to be
reasonable for use to trigger side-effects.
llvm-svn: 137823
-Wunused was a mistake. It resulted in duplicate warnings and lots of
other hacks. Instead, this should be a special sub-category to
-Wunused-value, much like -Wunused-result is.
Moved to -Wunused-comparison, moved the implementation to piggy back on
the -Wunused-value implementation instead of rolling its own, different
mechanism for catching all of the "interesting" statements.
I like the unused-value mechanism for this better, but its currently
missing several top-level statements. For now, I've FIXME-ed out those
test cases. I'll enhance the generic infrastructure to catch these
statements in a subsequent patch.
This patch also removes the cast-to-void fixit hint. This hint isn't
available on any of the other -Wunused-value diagnostics, and if we want
it to be, we should add it generically rather than in one specific case.
llvm-svn: 137822
code is very likely to be buggy, but its going to require more
significant changes on the part of the user to correct it in this case.
llvm-svn: 137820
a complement to the warnings we provide in condition expressions. Much
like we warn on conditions such as:
int x, y;
...
if (x = y) ... // Almost always a typo of '=='
This warning applies the complementary logic to "top-level" statements,
or statements whose value is not consumed or used in some way:
int x, y;
...
x == y; // Almost always a type for '='
We also mirror the '!=' vs. '|=' logic.
The warning is designed to fire even for overloaded operators for two reasons:
1) Especially in the presence of widespread templates that assume
operator== and operator!= perform the expected comparison operations,
it seems unreasonable to suppress warnings on the offchance that
a user has written a class that abuses these operators, embedding
side-effects or other magic within them.
2) There is a trivial source modification to silence the warning for
truly exceptional cases:
(void)(x == y); // No warning
A (greatly reduced) form of this warning has already caught a number of
bugs in our codebase, so there is precedent for it actually firing. That
said, its currently off by default, but enabled under -Wall.
There are several fixmes left here that I'm working on in follow-up
patches, including de-duplicating warnings from -Wunused, sharing code
with -Wunused's implementation (and creating a nice place to hook
diagnostics on "top-level" statements), and handling cases where a proxy
object with a bool conversion is returned, hiding the operation in the
cleanup AST nodes.
Suggestions for any of this code more than welcome. Also, I'd really
love suggestions for better naming than "top-level".
llvm-svn: 137819
e.g. for:
\define INVOKE(METHOD, CLASS) [CLASS METHOD]
void test2() {
INVOKE(meth, MyClass);
}
Pointing at 'meth' will give a CXCursor_ObjCMessageExpr and pointing at 'MyClass'
will give a CXCursor_ObjCClassRef.
llvm-svn: 137796
Example:
template <class T>
class A {
public:
template <class U> void f(U p) { }
template <> void f(int p) { } // <== class scope specialization
};
This extension is necessary to parse MSVC standard C++ headers, MFC and ATL code.
BTW, with this feature in, clang can parse (-fsyntax-only) all the MSVC 2010 standard header files without any error.
llvm-svn: 137573
Since -mkernel implies -fapple-kext, this just extends the current behavior
for -fapple-kext to apply for -mkernel as well. Radar 9933387.
llvm-svn: 137566
Report errors earlier: on checkDeadSymbols() and clear the state after the symbol we are tracking goes out of scope.
Also, perform lazy error checking. Instead of forcing the paths to be split depending one the return value of the allocator, make the return symbol depend on the allocated data symbol, which prolongs its life span to the time when the allocated data symbol becomes dead.
llvm-svn: 137523
Old warning:
warning: use of NULL in arithmetic operation [-Wnull-arithmetic]
return 10 <= NULL;
^ ~~~~
New warning:
warning: comparison between NULL and non-pointer ('int' and NULL) [-Wnull-arithmetic]
return 10 <= NULL;
~~ ^ ~~~~
llvm-svn: 137377
hierarchy of delegation, and that EH selector values are meaningful
function-wide (good thing, too, or inlining wouldn't work).
2,3d
1a
hierarchy of delegation and that EH selector values have the same
meaning everywhere in the function instead of being meaningful only
in the context of a specific selector.
This removes the need for routing edges through EH cleanups,
since a cleanup simply always branches to its enclosing scope.
llvm-svn: 137293
a member template, e.g.,
x.f<int>
if we have found a template in the type of x, but the lookup in the
current scope is ambiguous, just ignore the lookup in the current
scope. Fixes <rdar://problem/9915664>.
llvm-svn: 137255
overridden methods to diagnose their type mismatch.
This is a general solution for previous fixes
for // rdar://6191214 and // rdar://9352731
and removes lots of duplicate code.
llvm-svn: 137222
constructor. Previously, we did some bogus recursion into the fields
of anonymous structs (recursively), which ended up building invalid
ASTs that would cause CodeGen to crash due to invalid GEPs.
Now, we instead build the default initializations based on the
indirect field declarations at the top level, which properly generates
the sequence of GEPs needed to initialize the proper member. Fixes
PR10512 and <rdar://problem/9924046>.
llvm-svn: 137212
This patch special cases the parser for thread safety attributes so that all
attribute arguments are put in the argument list (instead of a special
parameter) since arguments may not otherwise resolve correctly without two-token
lookahead.
This patch also adds checks to make sure that attribute arguments are
lockable objects.
llvm-svn: 137130
situation. When a class explicitly or implicitly (through inheritance)
"conformsTo" two protocols which conflict (have methods which conflict).
This patch fixes the previous patch where warnings were coming out in
non-deterministic order. This is 2nd part of // rdar://6191214.
llvm-svn: 137055
ActOnStartOfSwitchStmt (i.e. before binding up a full-expression)
instead of ActOnFinishSwitchStmt.
Among other things, this means that property l-values are properly
converted inside the full-expression.
llvm-svn: 137014
aren't actually going to make a normal cleanup for. Sometimes
we optimistically create branches to such blocks for fixups,
and then we resolve the fixup to somewhere within the cleanup's
scope, and then the cleanup is actually not reachable for some
reason. The process of resolving the fixup leaves us with
switches whose default edge leads to the cleanup; we can
replace that with unreachable, then (in many cases) turn
the switch into an unconditional branch.
Fixes PR10467.
llvm-svn: 137011
case situations with the unary operators & and *. Also extend the array bounds
checking to work with pointer arithmetic; the pointer arithemtic checking can
be turned on using -Warray-bounds-pointer-arithmetic.
The changes to where CheckArrayAccess gets called is based on some trial &
error and a bunch of digging through source code and gdb backtraces in order
to have the check performed under as many situations as possible (such as for
variable initializers, arguments to function calls, and within conditional in
addition to the simpler cases of the operands to binary and unary operator)
while not being called--and triggering warnings--more than once for a given
ArraySubscriptExpr.
llvm-svn: 136997
This fixes cases where the anonymous bitfield is followed by a bitfield member.
E.g.,
struct t4
{
char foo;
long : 0;
char bar : 1;
};
rdar://9859156
llvm-svn: 136991
arrays. This now suppresses the warning only in the case of
a one-element array as the last field in a struct where the array size
is a literal '1' rather than any macro expansion or template parameter.
This doesn't distinguish between the language standard in use to allow
code which dates from C89 era to compile without the warning even in C99
and C++ builds. We could add a separate warning (under a different flag)
with fixit hints to switch to a flexible array, but its not clear that
this would be desirable. Much of the code using this idiom is striving
for maximum portability.
Tests were also fleshed out a bit, and the diagnostic itself tweaked to
be more pretty w.r.t. single elment arrays. This is more ugly than
I would like due to APInt's not being supported by the diagnostic
rendering engine.
A pseudo-patch for this was proposed by Nicola Gigante, but I reworked
it both for several correctness issues and for code style.
Sorry this was so long in coming.
llvm-svn: 136965
when performing typo correction involving any overloaded template functions.
The added test cases, while currently demontrating sub-optimal behavior, will
not trigger any messages without the 1-line change to SemaExpr.cpp.
llvm-svn: 136943
situation. When a class explicitly or implicitly (through inheritance)
"conformsTo" two protocols which conflict (have methods which conflict).
This is 2nd part of // rdar://6191214.
llvm-svn: 136927
Having a function declaration and definition with different types for a
parameter where the types have same (textual) name can occur when an unqualified
type name resolves to types in different namespaces in each location.
The error messages have been extended by adding notes that point to the first
parameter of the function definition that doesn't match the declaration, instead
of a generic "member declaration nearly matches". The generic message is still
used in cases where the mismatch is not in the paramenter list, such as
mismatched cv qualifiers on the member function itself.
llvm-svn: 136891
alignment. This fixes cases where the anonymous bitfield is followed by a
non-bitfield member. E.g.,
struct t4
{
int foo : 1;
long : 0;
char bar;
};
Part of rdar://9859156
llvm-svn: 136858
structures. Alignment can be enforced with the use of anonymous bitfields
(e.g., int :0), but this is not currently supported. Add this test case to
document the current state, which will hopefully be fixed shortly.
llvm-svn: 136848
designed to be executed, and its output inspected for correct values,
but we aren't executing it. We're just compiling it, and dumping it to
/dev/null. It also isn't freestanding. If there is a desire to have this
test actually stick around, complain and I'll revert this and try to add
the file checks necessary to make this actually test things.
llvm-svn: 136846
Change TypoCorrection to store a set of NamedDecls instead of a single
NamedDecl. Also add initial support for performing function overload
resolution to Sema::DiagnoseEmptyLookup.
llvm-svn: 136807
A homogeneous aggregate is an aggregate data structure where after flattening
any nesting there are 1 to 4 elements of the same base type that is either a
float, double, or Neon vector. All Neon vectors of the same size, either 64
or 128 bits, are treated as equivalent for this purpose. When using the
AAPCS-VFP ABI, check for homogeneous aggregates and pass them as arguments by
expanding them into a sequence of their base types. This requires extending
the existing support for expanded arguments to handle not only structs, but
also constant arrays and complex types.
llvm-svn: 136767
1. Be more tolerant of comments in -CC (comment-preserving) mode. We were missing a few cases.
2. Make sure to expand the second FOO in "#if defined FOO FOO". (See also
r97253, which addressed the case of "#if defined(FOO FOO".)
Fixes PR10286.
llvm-svn: 136748
has a single element. This disables the warning in cases where
there is a clear bug, but this is really rare (who uses arrays
with one element?) and it also silences a large class of false
positive issues with C89 code that is using tail padding in structs.
A better version of this patch would detect when an array is in
a tail position in a struct, but at least patch fixes the huge
false positives that are hitting postgres and other code.
llvm-svn: 136724
by eliminating the type ID from constructor, destructor, and
conversion function names. There are several reasons for this change:
- A given type (say, int*) isn't guaranteed to have a single, unique
type ID within a chain of PCH files. Hence, we could end up hashing
based on the wrong type ID, causing name lookup to fail.
- The mapping from types back to type IDs required one DenseMap
entry for every type that was ever deserialized, which was an
unacceptable cost to support just the name lookup of constructors,
destructors, and conversion functions. Plus, this mapping could
never actually work with chained or multiple PCH, based on the first
bullet.
Once we have eliminated the type from the hash function, these
problems go away, as does my horrible "reverse type remap" hack, which
was doomed from the start (see bullet #1 above) and far too
complicated.
However, note that removing the type from the hash function means that
all constructors, destructors, and conversion functions have the same
hash key, so I've updated the caller to double-check that the
declarations found have the appropriate name.
llvm-svn: 136708
broken because the end location of the parameter was the end location of the default arg,
resulting in a source range that could begin in one file and end in another.
llvm-svn: 136572
appropriate sentinel at the end of the argument list. Also, put the
sentinel warnings under -Wsentinel. Fixes <rdar://problem/8764236>.
llvm-svn: 136566
variadic bit (", ..." or ", args...") into the prior placeholder, like
we do with functions and methods. Fixes <rdar://problem/9740808>.
llvm-svn: 136563
already-defined and forward-declared results. Already-defined results
are fine because they could be the start of a category. Fixes
<rdar://problem/9811691>.
llvm-svn: 136559
special diagnostic for ARC ownership-qualified types. We wouldn't want
to expose Objective-C programmers to the term "POD", would we? Fixes
<rdar://problem/9772982>.
llvm-svn: 136558
Fariborz Jahanian