Richard Smith [Fri, 27 Jul 2012 04:22:15 +0000 (04:22 +0000)]
Final piece of core issue 1330: delay computing the exception specification of
a defaulted special member function until the exception specification is needed
(using the same criteria used for the delayed instantiation of exception
specifications for function temploids).
EST_Delayed is now EST_Unevaluated (using 1330's terminology), and, like
EST_Uninstantiated, carries a pointer to the FunctionDecl which will be used to
resolve the exception specification.
This is enabled for all C++ modes: it's a little faster in the case where the
exception specification isn't used, allows our C++11-in-C++98 extensions to
work, and is still correct for C++98, since in that mode the computation of the
exception specification can't fail.
The diagnostics here aren't great (in particular, we should include implicit
evaluation of exception specifications for defaulted special members in the
template instantiation backtraces), but they're not much worse than before.
Our approach to the problem of cycles between in-class initializers and the
exception specification for a defaulted default constructor is modified a
little by this change -- we now reject any odr-use of a defaulted default
constructor if that constructor uses an in-class initializer and the use is in
an in-class initialzer which is declared lexically earlier. This is a closer
approximation to the current draft solution in core issue 1351, but isn't an
exact match (but the current draft wording isn't reasonable, so that's to be
expected).
Jordan Rose [Fri, 27 Jul 2012 01:15:02 +0000 (01:15 +0000)]
[analyzer] Look through SubstNonTypeTemplateParmExprs.
We were treating this like a CXXDefaultArgExpr, but
SubstNonTypeTemplateParmExpr actually appears when a template is
instantiated, i.e. we have all the information necessary to evaluate it.
This allows us to inline functions like llvm::array_lengthof.
Jordan Rose [Fri, 27 Jul 2012 00:47:52 +0000 (00:47 +0000)]
[analyzer] Use a stack-based local AGAIN to fix the build for real.
It's a good thing CallEvents aren't created all over the place yet.
I checked all the uses this time and the private copy constructor
/really/ shouldn't cause any more problems.
objc-arc: When objects with known CF semantics are assigned to
retainable types in arc, only suggest CFBridgingRelease/
CFBridgingRetain and not the confusing __bridge casts.
// rdar://11923822
Look at the preceding CFGBlock for the expression to load from in ExprEngine::VisitGuardedExpr
instead of walking to the preceding PostStmt node. There are cases where the last evaluated
expression does not appear in the ExplodedGraph.
Jordan Rose [Thu, 26 Jul 2012 21:41:15 +0000 (21:41 +0000)]
[analyzer] CallEvent is no longer a value object.
After discussion, the type-based dispatch was decided to be bad for
maintenance and made it very easy for subtle bugs to creep in. Instead,
we'll just be very careful when we do have to allocate these on the heap.
Jordan Rose [Thu, 26 Jul 2012 20:04:30 +0000 (20:04 +0000)]
[analyzer] Don't crash on implicit statements inside initializers.
Our BugReporter knows how to deal with implicit statements: it looks in
the ParentMap until it finds a parent with a valid location. However, since
initializers are not in the body of a constructor, their sub-expressions are
not in the ParentMap. That was easy enough to fix in AnalysisDeclContext.
...and then even once THAT was fixed, there's still an extra funny case
of Objective-C object pointer fields under ARC, which are initialized with
a top-level ImplicitValueInitExpr. To catch these cases,
PathDiagnosticLocation will now fall back to the start of the current
function if it can't find any other valid SourceLocations. This isn't great,
but it's miles better than a crash.
(All of this is only relevant when constructors and destructors are being
inlined, i.e. under -cfg-add-initializers and -cfg-add-implicit-dtors.)
Jordan Rose [Thu, 26 Jul 2012 20:04:25 +0000 (20:04 +0000)]
[analyzer] Don't crash on array constructors and destructors.
This workaround is fairly lame: we simulate the first element's constructor
and destructor and rely on the region invalidation to "initialize" the rest
of the elements.
Jordan Rose [Thu, 26 Jul 2012 20:04:13 +0000 (20:04 +0000)]
[analyzer] Handle base class initializers and destructors.
Most of the logic here is fairly simple; the interesting thing is that
we now distinguish complete constructors from base or delegate constructors.
We also make sure to cast to the base class before evaluating a constructor
or destructor, since non-virtual base classes may behave differently.
This includes some refactoring of VisitCXXConstructExpr and VisitCXXDestructor
in order to keep ExprEngine.cpp as clean as possible (leaving the details for
ExprEngineCXX.cpp).
Jordan Rose [Thu, 26 Jul 2012 20:04:05 +0000 (20:04 +0000)]
[analyzer] Show paths for destructor calls.
This modifies BugReporter and friends to handle CallEnter and CallExitEnd
program points that came from implicit call CFG nodes (read: destructors).
This required some extra handling for nested implicit calls. For example,
the added multiple-inheritance test case has a call graph that looks like this:
In this case we correctly notice that we started in an inlined function
when we reach the CallEnter program point for the second ~SmartPointer.
However, when we reach the next CallEnter (for ~Subclass), we were
accidentally re-using the inner ~SmartPointer call in the diagnostics.
Rather than guess if we saw the corresponding CallExitEnd based on the
contents of the active path, we now just ask the PathDiagnostic if there's
any known stack before popping off the top path.
(A similar issue could have occured without multiple inheritance, but there
wasn't a test case for it.)
Anna Zaks [Thu, 26 Jul 2012 00:27:51 +0000 (00:27 +0000)]
[analyzer] Inline ObjC class methods.
- Some cleanup(the TODOs) will be done after ObjC method inlining is
complete.
- Simplified CallEvent::getDefinition not to require ISDynamicDispatch
parameter.
- Also addressed Jordan's comments from r160530.
Update ExprEngine's handling of ternary operators to find the ternary expression
value by scanning the path, rather than assuming we have visited the '?:' operator
as a terminator (which sets a value indicating which expression to grab the
final ternary expression value from).
objective-c modern translator. Provide declaration of
"memset' lazily when is needed in translation of
struct-valued methods which require checkinf of nil receivers
outside their bodies. // rdar://11847319
Remove experimental invalid iterators checker from the codebase until we have the time
to fix all the issues. Currently the code is essentially unmaintained and buggy, and
needs major revision (with coupled enhancements to the analyzer core).
Chad Rosier [Wed, 25 Jul 2012 17:52:16 +0000 (17:52 +0000)]
[driver crash diagnostics] A few enhancements:
-Strip -iquote and -M options.
-Quote -D options to avoid problems with command line macros that include
parens.
rdar://11949066
Richard Smith [Wed, 25 Jul 2012 03:56:55 +0000 (03:56 +0000)]
PR12057: Allow variadic template pack expansions to cross lambda boundaries.
Rather than adding a ContainsUnexpandedParameterPack bit to essentially every
AST node, we tunnel the bit directly up to the surrounding lambda expression
when we reach a context where an unexpanded pack can not normally appear.
Thus any statement or declaration within a lambda can now potentially contain
an unexpanded parameter pack.
This was actually supported in Clang 3.1, but we got a request for a
__has_feature so that header files can use this more safely. It's
unfortunate that the 3.1 release doesn't include this, however.
Richard Smith [Tue, 24 Jul 2012 21:02:14 +0000 (21:02 +0000)]
When a && or || appears as the condition of a ?:, perform appropriate
short-circuiting when building the CFG. Also be sure to skip parens before
checking for the && / || special cases. Finally, fix some crashes in CFG
printing in the presence of calls to destructors for array of array of class
type.
Dmitri Gribenko [Tue, 24 Jul 2012 17:52:18 +0000 (17:52 +0000)]
Comment parsing: couple TextTokenRetokenizer and comment parser together to
remove one of the two variable-length lookahead buffers. Now retokenizer will
ask for more tokens when it needs them.
Manuel Klimek [Tue, 24 Jul 2012 13:37:29 +0000 (13:37 +0000)]
Introduces a new concept for binding results to matchers
as per Chandler's request:
- introduces a new matcher base type BindableMatcher that
provides the bind() call
- makes all dynamic-cast matcher creation functions return
BindableMatchers; the special case about dynamic-cast
matchers is that the node they match on and the node
their child matchers match on are the same node, just
casted to a different type; thus, there is no ambiguity
on what bind() matches on; additionally, those are the
matchers that we name with nouns in the matcher language,
so it's easy for users to intuitively know which matchers
are bindable
To make this change possible, we got rid of a non-orthogonal
implementation of thisPointerType, which had an implicit
dynamic-cast matcher from CallExpr to CXXMemberCallExpr; as
alternative, we now provide a memberCall dynamic-cast matcher
and thisPointerType is a predicate on CXXMemberCallExpr.
Last, the ArgumentAdaptingMatcher is actually not required
for the implementation of makeDynCastAllOfComposite - this
simplification makes it more obvious where the bind() call
can be used based on the matcher creation function types.
Nick Lewycky [Tue, 24 Jul 2012 01:40:49 +0000 (01:40 +0000)]
Emit debug info for dynamic initializers. Permit __attribute__((nodebug)) on
variables that have static storage duration, it removes debug info on the
emitted initializer function but not all debug info about this variable.
Dmitri Gribenko [Mon, 23 Jul 2012 19:41:49 +0000 (19:41 +0000)]
libclang comments AST: clang_ParamCommandComment_getParamName: don't assert
when a \param command does not have a parameter name, just return an empty
string instead.
Michael Han [Mon, 23 Jul 2012 18:48:41 +0000 (18:48 +0000)]
Refactor handler functions for thread safety attributes.
Make handler functions for thread safety attributes consistent with other attributes handler functions
by removing the bool parameter from some of the thread safety attributes handler functions and extracting
common checks out of different handler functions.
Richard Smith [Mon, 23 Jul 2012 05:45:25 +0000 (05:45 +0000)]
Add diagnostics for comma at end of enum and for extra semicolon at namespace
scope to -Wc++11-extensions. Move extra semicolon after member function
definition diagnostic out of -pedantic, since C++ allows a single semicolon
there. Keep it in -Wextra-semi, though, since it's still questionable.
Douglas Gregor [Mon, 23 Jul 2012 04:23:39 +0000 (04:23 +0000)]
When we have an Objective-C object with non-trivial lifetime in a
structor class under ARC, that struct/class does not have a trivial
move constructor or move assignment operator. Fixes the rest of
<rdar://problem/11738725>.
Richard Smith [Sun, 22 Jul 2012 23:45:10 +0000 (23:45 +0000)]
PR12917: Remove incorrect assumption that lambda mangling information cannot
change once it's been assigned. It can change in two ways:
1) In a template instantiation, the context declaration should be the
instantiated declaration, not the declaration in the template.
2) If a lambda appears in the pattern of a variadic pack expansion, the
mangling number will depend on the pack length.
Dmitri Gribenko [Sat, 21 Jul 2012 01:47:43 +0000 (01:47 +0000)]
Comment to HTML conversion: add more CSS classes to identify function arguments
by index. This is useful if the user does not document all arguments, and we
can't find a particular argument by index via :nth-of-type() CSS selector.
projects / clang / log