Set invalid type of declarator after emitting error diagnostics,
so that it won't be later considered when instantiating the template.
Added test5_inst in test/SemaCXX/condition.cpp for non-regression.
llvm-svn: 172201
It's better to show no caret/ranges/fixits than to show them misaligned.
(It scares me a bit that Clang is going to show them misaligned on
systems without a proper llvm::sys::locale::columnWidth.) See PR14910.
This reverts commit a03cc72bedd015b7eb76e9ac650992f58a48549d.
llvm-svn: 172163
assertions.
To ensure that custom assertions/conditional would also be supported,
just check if the ivar that needs to be invalidated or set to nil is
compared against 0.
Unfortunately, this will not work for code containing 'assert(IvarName)'
llvm-svn: 172147
storage and thus is implicitly zero-initialized, no need to
do C++11 memory model. This patch unconditionally detects
such condition and zeroinitializer's the variable.
Patch has been commented on and OKed by Doug off-line.
// rdar://12897704
llvm-svn: 172144
In some cases, we just pick any ivar that needs invalidation and attach
the warning to it. Picking the first from DenseMap of pointer keys was
triggering non-deterministic output.
llvm-svn: 172134
Restructured the checker so that it could easily find two new classes of
issues:
- when a class contains an invalidatable ivar, but no declaration of an
invalidation method
- when a class contains an invalidatable ivar, but no definition of an
invalidation method in the @implementation.
The second case might trigger some false positives, for example, when
the method is defined in a category.
llvm-svn: 172104
Don't do this in Google style though:
http://google-styleguide.googlecode.com/svn/trunk/objcguide.xml#Protocols
Most other places (function declarations, variable declarations) still get
this wrong, and since this looks very similiar to template instantiations to
the lexer (`id <MyProtocol> a = ...`), it's going to be hard to fix in some
places.
llvm-svn: 172099
The first token in @implementation, @interface, and @protocol lines is now
marked TT_ObjCDecl, and lines starting with a TT_ObjCDecl token are now marked
LT_ObjCMethodDecl.
llvm-svn: 172093
Not only is this inefficient for TableGen, it's annoying for maintenance
when renaming warning flags (unusual) or adding those flags to a group
(more likely).
This uses the new fix-it infrastructure for LLVM's SourceMgr/SMDiagnostic,
as well as a few changes to TableGen to track more source information.
llvm-svn: 172087
Before: int (^myBlock) (int) = ^(int num) {}
A<void ()>;
int (*b)(int);
After: int (^myBlock)(int) = ^(int num) {}
A<void()>;
int(*b)(int);
For function types and function pointer types, this patch only makes
the behavior consistent (for types that are keywords and other types).
For the latter function pointer type declarations, we'll probably
want to add a space after "int".
Also added LangOpts.Bool = 1, so we handle "A<bool()>" appropriately
Moved the LangOpts-settings to a public place for use by tests
and clang-format binary.
llvm-svn: 172065
Just reuse the @interface code for this. It accepts slightly more than
necessary (@implementation cannot have protocol lists), but that's ok.
llvm-svn: 172019
there are macro expansions inside macro arguments where the arguments are
not expanded in the same order as listed; don't assert that all macro expansions
are in source order.
rdar://12397063
llvm-svn: 172018
in case condition type. // rdar://11577384.
Test is conditionalized on x86_64-apple triple as
I am not sure if the INT_MAX/LONG_MAX values in the test
will pass this test for other hosts.
llvm-svn: 172016
(because they are part of some module) but have not been made visible
(because they are in a submodule that wasn't imported), filter out
those declarations unless both the old declaration and the new
declaration have external linkage. When one or both has internal
linkage, there should be no conflict unless both are imported.
llvm-svn: 171925
In the source
static void f();
static void f();
template<typename T>
static void g() {
f();
}
static void f() {
}
void h() {
g<int>();
}
the call to f refers to the second decl, but it is only marked used at the end
of the translation unit during instantiation, after the third f decl has been
linked in.
With this patch we mark all subsequent decls used, so that it is easy to check
if a symbol is used or not.
llvm-svn: 171888
The issue here is that if we have 2 leaks reported at the same line for
which we cannot print the corresponding region info, they will get
treated as the same by issue_hash+description. We need to AUGMENT the
issue_hash with the allocation info to differentiate the two issues.
Add the "hash" (offset from the beginning of a function) representing
allocation site to solve the issue.
We might want to generalize solution in the future when we decide to
track more than just the 2 locations from the diagnostics.
llvm-svn: 171825
The test should be looking for gcrt0.o not crt0.o. Clang was already printing
"gcrt0", but the test was looking for "{{.*}}crt0.o", and the .* regexp
consumed "g".
Patch by Brad Smith.
llvm-svn: 171815
DenseMap and the lookups associated with it.
Instead of lexing for preprocessor tokens, associating the annotation with the location in a map,
and later lookup in the map in order to adjust the cursor annotation, just annotate the cursor
while lexing for preprocessor tokens.
llvm-svn: 171775
Previously type/storage qualifiers would not be annotated as the declaration they belonged to.
Just use the resulting source range of getRawCursorExtent() which is more correct
than what AnnotateTokensWorker::Visit() was adjusting it to.
llvm-svn: 171774
this was ever a macro name and return a specific CXCursor_MacroExpansion cursor in such a case,
instead of the generic CXCursor_MacroDefinition.
Checking for macro name makes sure the identifier is not part of the identifier list in a
function macro.
While, in general, resolving identifiers in macro definitions to other macros may not be completely accurate,
it greatly improves functionality such as give-me-the-definition-of-this, which was not working at all
inside macro definitions.
llvm-svn: 171773
Better handle the blacklisting of known bad deallocators when symbol
escapes through a call to CFStringCreateWithBytesNoCopy.
Addresses radar://12702952.
llvm-svn: 171770
When a property is "inherited" through both a parent class and directly
through a protocol, we should not require the child to invalidate it
since the backing ivar belongs to the parent class.
(Fixes radar://12913734)
llvm-svn: 171769
Previous indent:
class A {
}
a;
void f() {
};
With this patch:
class A {
} a;
void f() {
}
;
The patch introduces a production for classes and structs, and parses
the rest of the line to the semicolon after the class scope.
This allowed us to remove a long-standing wart in the parser that would
just much the semicolon after any block.
Due to this suboptimal formating some tests were broken.
Some unrelated formatting tests broke; those hit a bug in the ast
printing, and need to be fixed separately.
llvm-svn: 171761
Following r168626, in class declaration or definition, there are a combination of syntactic locations
where C++11 attributes could appear, and among those the only valid location permitted by standard is
between class-key and class-name. So for those attributes appear at wrong locations, fixit is used to
move them to expected location and we recover by applying them to the class specifier.
llvm-svn: 171757
with respect to the lower "left-hand-side bitwidth" bits, even when negative);
see OpenCL spec 6.3j. This patch both implements this behaviour in the code
generator and "constant folding" bits of Sema, and also prevents tests
to detect undefinedness in terms of the weaker C99 or C++ specifications
from being applied.
llvm-svn: 171755
First check only wrapped with i==8, second wrapped at i==2,8,18,28,...
This fix restores the intended behavior: i==8,18,28,...
Found with -fsanitize=integer.
llvm-svn: 171718
Using added LLVM functionality in r171698. This works in GDB for member
variable pointers but not member function pointers. See the LLVM commit and
GDB bug 14998 for details.
Un-xfailing cases in the GDB 7.5 test suite will follow.
llvm-svn: 171699
Catch some cases I'd missed in r171605 related to unnamed parameters of record
type. This resolves all remaining cases of PR14573 suppression in the GDB 7.5
test suite. Fix to the test suite to follow.
llvm-svn: 171633
Referring back to the original commit (r115090) which was a frontend only test
I adjusted this test to verify the frontend change that was made, to emit the
protected access value in the flags metadata field.
llvm-svn: 171604
It is somewhat hard to test linkage, so I decided to try to add an assert. This
already found some interesting cases where there were different.
llvm-svn: 171585
Some of this is still pretty rough (note the load of FIXMEs), but it is
strictly an improvement and fixes various bugs that were related to
macro processing but are also imporant in non-macro use cases.
Specific fixes:
- correctly puts espaced newlines at the end of the line
- fixes counting of white space before a token when escaped newlines are
present
- fixes parsing of "trailing" tokens when eof() is hit
- puts macro parsing orthogonal to parsing other structure
- general support for parsing of macro definitions
Due to the fix to format trailing tokens, this change also includes a
bunch of fixes to the c-index tests.
llvm-svn: 171556
__strong __block variables, perform objc_storeStrong on
source and destination instead of direct move. This
is done with -O0 and to improve some analysis.
// rdar://12530881
llvm-svn: 171555
The line information was changed when emitting debug information for all the
DeclRefExprs and we should change it back to get ready for PopClenupBlocks
called from FinishFunction.
rdar://11562117
llvm-svn: 171493
To do so we have to wait until we know that the type of a variable has been
deduced. Sema::FinalizeDeclaration is the first callback that is used for
decl with or without initializers.
llvm-svn: 171458
This fixes pr14736. It is fairly ugly, but I don't think we can do much better
as we have to wait at least until the end of the typedef to know if the
function will have external linkage or not.
llvm-svn: 171240
This patch moves hasCLanguageLinkage to be VarDecl and FunctionDecl methods
so that they can be used from SemaOverload.cpp and then fixes the logic
in Sema::IsOverload.
llvm-svn: 171193
This makes the test not dependent on LLVM & won't vary/break based on LLVM
codegen related changes. Appropriately testing at the Clang level what was
fixed at the Clang level originally (in r124210).
llvm-svn: 171175
the body of a functions. The problem was that hasBody looks at the entire chain
and causes problems to -fvisibility-inlines-hidden if the cache was not
invalidated.
Original message:
Cache visibility of decls.
This unifies the linkage and visibility caching. I first implemented this when
working on pr13844, but the previous fixes removed the performance advantage of
this one.
This is still a step in the right direction for making linkage and visibility
cheap to use.
llvm-svn: 171053
Using the file immediately after "> file 2>&1" causes weird behavior on win32.
For example, "foo > %t 2>&1; FileCheck --input-file=%t"
Use "foo 2>&1 | tee %t" instead.
Note, mixing stdout and stderr to the same handle might cause unexpected behavior, due to buffering.
llvm-svn: 171009
code-completion results, the SourceManager state may be slightly
different when code-completing.
And we don't even care for diagnostics when code-completing, anyway.
llvm-svn: 170979
This was removed with -Wunique-enum, which is still removed. The
corresponding thread on cfe-comments for that warning is here:
http://lists.cs.uiuc.edu/pipermail/cfe-dev/2012-September/024224.html
If we get specific user feedback for -Wduplicate-enum we can evaluate
whether or not to keep it.
llvm-svn: 170974
Our internal buildbot just failed because the tempfile happened to be named
'nodefaultlib-lcGA7k.o', which contains the substring '-lc'.
llvm-svn: 170941
Along the way, fix a bug in CheckLiteralKind(), previously in diagnoseObjCLiteralComparison, where we didn't ignore parentheses
in boxed expressions for purpose of classification.
In other words, both @42 and @(42) should be classified as numeric
literals.
llvm-svn: 170931
found by running -ast-print on all-std-headers.cpp
which caused it to go into infinite loop. Now
-ast-print prints all declarations found in
all-std-headers.cpp.
llvm-svn: 170928
deterministic.
Commit message for r170826:
[analyzer] Traverse the Call Graph in topological order.
Modify the call graph by removing the parentless nodes. Instead all
nodes are children of root to ensure they are all reachable. Remove the
tracking of nodes that are "top level" or global. This information is
not used and can be obtained from the Decls stored inside
CallGraphNodes.
Instead of existing ordering hacks, analyze the functions in topological
order over the Call Graph.
Together with the addition of devirtualizable ObjC message sends and
blocks to the call graph, this gives around 6% performance improvement
on several large ObjC benchmarks.
llvm-svn: 170906
CXXScalarValueInitExpr (or an ImplicitValueInitExpr), strip it back down to an
empty pair of parentheses so that the initialization code can tell that we're
performing value-initialization.
llvm-svn: 170867
When we are visiting the extern declaration of 'i' in
static int i = 99;
int foo() {
extern int i;
return i;
}
We should not try to handle it as if it was an function static. That is, we
must consider the written storage class.
Fixing this then exposes that the assert in EmitGlobalVarDeclLValue and the
if leading to its call are not completely accurate. They were passing before
because the second decl was marked as having external storage. I changed them
to check the linkage, which I find easier to understand.
Last but not least, there is something strange going on with cuda and opencl.
My guess is that the linkage computation for these languages needs to be
audited, but I didn't want to change that in this patch so I just updated
the storage classes to keep the current behavior.
Thanks to Reed Kotler for reporting this.
llvm-svn: 170827
Modify the call graph by removing the parentless nodes. Instead all
nodes are children of root to ensure they are all reachable. Remove the
tracking of nodes that are "top level" or global. This information is
not used and can be obtained from the Decls stored inside
CallGraphNodes.
Instead of existing ordering hacks, analyze the functions in topological
order over the Call Graph.
Together with the addition of devirtualizable ObjC message sends and
blocks to the call graph, this gives around 6% performance improvement
on several large ObjC benchmarks.
llvm-svn: 170826
to also remove a trailing space if possible.
For example, removing '__bridge' from:
i = (__bridge I*)p;
should result in:
i = (I*)p;
not:
i = ( I*)p;
rdar://11314821
llvm-svn: 170764
produce a note for that diagnostic either with a different DiagnosticEngine or
after calling DiagnosticEngine::Reset(). That didn't make any sense, and did the
wrong thing if the original diagnostic was suppressed.
llvm-svn: 170636
modules when getting the decls for a namespace or translation unit.
Otherwise the code-completion results will not be complete.
rdar://12889089
llvm-svn: 170596
too). When instantiating a direct-initializer, if we find it has zero
arguments, produce an empty ParenListExpr rather than returning a null
expression.
llvm-svn: 170490
copy-list-initialization (and doesn't add an additional copy step):
Fill in the ListInitialization bit when creating a CXXConstructExpr. Use it
when instantiating initializers in order to correctly handle instantiation of
copy-list-initialization. Teach TreeTransform that function arguments are
initializations, and so need this special treatment too. Finally, remove some
hacks which were working around SubstInitializer's shortcomings.
llvm-svn: 170489
This fixes a subtle bug reported in <rdar://problem/12584554> where a double-nested
macro could lead to an incorrect fixit location with live issues.
This fix also uncovers a bunch of subtle bugs in our indexer test cases which
are now fixed (mostly around source ranges for attributes).
llvm-svn: 170468
use clang's formatter. Currently, formatter is used
to format declaration tags for xml comments. Since formatter
is in flux and its change will break several of the clang comment
tests, only a single tests is formatted using this facility.
Doug has reviewed and approved it for check-in.
llvm-svn: 170467
PR 14529 was opened because neither Clang or LLVM was expanding
calls to creal* or cimag* into instructions that just load the
respective complex field. After some discussion, it was not
considered realistic to do this in LLVM because of the platform
specific way complex types are expanded. Thus a way to solve
this in Clang was pursued. GCC does a similar expansion.
This patch adds the feature to Clang by making the creal* and
cimag* functions library builtins and modifying the builtin code
generator to look for the new builtin types.
llvm-svn: 170455
This fixes the storage class of extern decls that are merged with file level
statics. The patch also fixes the linkage computation so that they are
considered internal.
llvm-svn: 170406
This also requires adding support to -cc1as for passing the detecting
PWD down through LLVM's debug info (which in turn required the LLVM
change in r170371).
The test case is weak (we only test the driver behavior) because there
is currently to infrastructure for running cc1as in the test suite. So
those four lines are untested (much like all other lines in that file),
but we have a test for the same pattern using llvm-mc in the LLVM
repository.
llvm-svn: 170373
performance heuristic
After inlining a function with more than 13 basic blocks 32 times, we
are not going to inline it anymore. The idea is that inlining large
functions leads to drastic performance implications. Since the function
has already been inlined, we know that we've analyzed it in many
contexts.
The following metrics are used:
- Large function is a function with more than 13 basic blocks (we
should switch to another metric, like cyclomatic complexity)
- We consider that we've inlined a function many times if it's been
inlined 32 times. This number is configurable with -analyzer-config
max-times-inline-large=xx
This heuristic addresses a performance regression introduced with
inlining on one benchmark. The analyzer on this benchmark became 60
times slower with inlining turned on. The heuristic allows us to analyze
it in 24% of the time. The performance improvements on the other
benchmarks I've tested with are much lower - under 10%, which is
expected.
llvm-svn: 170361
Fariborz Jahanian