Add some tests for __OPTIMIZE_SIZE__ and __NO_INLINE__,
removing the superfluous copies in the target-specific
tests, since it's target-independent.
This uncovered a bug in the handling of -Oz: it would
attempt to store the value 2 in the 1-bit bitfield OptimizeSize,
leaving a value of 0 and never defining __OPTIMIZE_SIZE__.
Richard Smith [Wed, 8 Aug 2012 06:13:49 +0000 (06:13 +0000)]
Implement final piece of DR963 and also DR587:
A conditional operator between glvalues of types cv1 T and cv2 T produces a
glvalue if the expressions are of the same value kind and one of cv1 and cv2
is a subset of the other.
A conditional operator between two null pointer constants is permitted if one
of them is of type std::nullptr_t.
Eli Friedman [Wed, 8 Aug 2012 04:39:56 +0000 (04:39 +0000)]
Get rid of an early return in Sema::ActOnFields which doesn't make sense anymore.
Fixes a crash (<rdar://problem/11067144>), and generally seems to improve
recovery in other cases.
Richard Smith [Tue, 7 Aug 2012 22:06:48 +0000 (22:06 +0000)]
When building a conditional operator where one operand is a throw-expression
and the other is a glvalue of class type, don't forget to copy-initialize a
temporary when performing the lvalue-to-rvalue conversion on the glvalue.
Strangely, DefaultLvalueConversions misses this part of the lvalue-to-rvalue
conversions.
Richard Smith [Tue, 7 Aug 2012 21:30:42 +0000 (21:30 +0000)]
-Wunused-private-fields: Don't try to check unresolved initializer expressions
for side-effects. Instead, check for side-effects after performing
initialization. Doing so also removes some strange corner cases and differences
between in-class initialization and constructor initialization.
Anna Zaks [Tue, 7 Aug 2012 18:36:58 +0000 (18:36 +0000)]
Turn on strncat-size warning implemented a while ago.
Warns on anti-patterns/typos in the 'size' argument to strncat. The
correct size argument should look like the following:
- strncat(dst, src, sizeof(dst) - strlen(dest) - 1);
We warn on:
- sizeof(dst)
- sizeof(src)
- sizeof(dst) - strlen(dst)
- sizeof(src) - anything
(This has been implemented in void Sema::CheckStrncatArguments().)
Dmitri Gribenko [Tue, 7 Aug 2012 17:54:38 +0000 (17:54 +0000)]
libclang API for comment-to-xml conversion.
The implementation also includes a Relax NG schema and tests for the schema
itself. The schema is used in c-index-test to verify that XML documents we
produce are valid. In order to do the validation, we add an optional libxml2
dependency for c-index-test.
Credits for CMake part go to Doug Gregor. Credits for Autoconf part go to Eric
Christopher. Thanks!
Richard Smith [Tue, 7 Aug 2012 05:18:29 +0000 (05:18 +0000)]
Update documentation of HasSideEffects to match its callers' expectations, and
update implementation to match. An elidable, non-trivial constructor call is a
side-effect under this definition, but wasn't under the old one, because we are
not required to evaluate it even though it may have an effect.
Also rationalize checking for volatile reads: just look for lvalue-to-rvalue
conversions on volatile glvalues, and ignore whether a DeclRefExpr etc is for
a volatile variable.
Ted Kremenek [Tue, 7 Aug 2012 05:01:49 +0000 (05:01 +0000)]
Enhance ClangDiagnosticsEmitter to reject diagnostics that are errors that are also
included in warning groups. Warning groups can only contain warnings, because only
warnings can be mapped to errors or ignored.
This caught a few diagnostics that were incorrectly in diagnostic groups, and
could have resulted in a compiler crash when those diagnostic groups were mapped.
Richard Smith [Tue, 7 Aug 2012 04:16:51 +0000 (04:16 +0000)]
Teach Expr::HasSideEffects about all the Expr types, and fix a bug where it
was mistakenly classifying dynamic_casts which might throw as having no side
effects.
Switch it from a visitor to a switch, so it is kept up-to-date as future Expr
nodes are added. Move it from ExprConstant.cpp to Expr.cpp, since it's not
really related to constant expression evaluation.
Since we use HasSideEffect to determine whether to emit an unused global with
internal linkage, this has the effect of suppressing emission of globals in
some cases.
I've left many of the Objective-C cases conservatively assuming that the
expression has side-effects. I'll leave it to someone with better knowledge
of Objective-C than mine to improve them.
Ted Kremenek [Tue, 7 Aug 2012 00:34:48 +0000 (00:34 +0000)]
Remove two fall through diagnostics from the IgnoredAttributes group, since they are errors, not warnings.
Only warnings should be in warning groups. This was causing an assertion failure in
code using pragmas to map -Wignored-attributes to ignored, but this kind of
diagnostic mapping is disallowed for builtin errors. We should just wire
up tablegen to reject errors being in diagnostic groups.
Chad Rosier [Tue, 7 Aug 2012 00:29:06 +0000 (00:29 +0000)]
[ms-inline asm] Stmt destructors are never called, so allocate the AsmToks using
the ASTContext BumpPtr. Also use the preferred llvm::ArrayRef interface.
Anna Zaks [Mon, 6 Aug 2012 23:25:45 +0000 (23:25 +0000)]
[analyzer] Dynamic type info - propagate through implicit casts.
I currently have a bit of redundancy with the cast kind switch statement
inside the ImplicitCast callback, but I might be adding more casts going
forward.
Anna Zaks [Mon, 6 Aug 2012 23:25:39 +0000 (23:25 +0000)]
[analyzer] Add a checker to manage dynamic type propagation.
Instead of sprinkling dynamic type info propagation throughout
ExprEngine, the added checker would add the more precise type
information on known APIs (Ex: ObjC alloc, new) and propagate
the type info in other cases (ex: ObjC init method, casts (the second is
not implemented yet)).
Add handling of ObjC alloc, new and init to the checker.
Jordan Rose [Mon, 6 Aug 2012 21:28:14 +0000 (21:28 +0000)]
[analyzer] Add plist output checks for all four "path notes" tests.
No functionality change, but from now on, any new path notes should be
tested both with plain-text output (for ease of human auditing) and with
plist output (to ensure control flow and events are being correctly
represented in Xcode).
Jordan Rose [Mon, 6 Aug 2012 21:28:02 +0000 (21:28 +0000)]
[analyzer] Ignore OS X 10.8's annotations for NSMakeCollectable.
The frameworks correctly use the 'cf_consumed' and 'ns_returns_retained'
attributes for NSMakeCollectable, but we can model the behavior under
garbage collection more precisely than that.
Richard Trieu [Mon, 6 Aug 2012 21:09:23 +0000 (21:09 +0000)]
For global record types, the self reference checker was called twice, resulting
in duplicate -Wuninitialized warnings. Change so that only the check in
TryConstructorInitialization() will be used and a single warning be emitted.
Chad Rosier [Mon, 6 Aug 2012 20:03:45 +0000 (20:03 +0000)]
[ms-inline asm] Pass Tokens to Sema and store them in the AST. No functional
change intended. No test case as there's no real way to test at this time.
Dmitri Gribenko [Mon, 6 Aug 2012 17:08:27 +0000 (17:08 +0000)]
Comment diagnostics: warn on duplicate \brief and \return commands.
Doxygen manual claims that multiple \brief or \returns commands will be merged
together, but actual behavior is different (second \brief command becomes a
part of a discussion, second \returns becomes a "Returns: blah" paragraph on
its own). Anyway, it seems to be a bad idea to use multiple \brief or \returns
commands in a single command.
Richard Smith [Mon, 6 Aug 2012 04:09:06 +0000 (04:09 +0000)]
PR13529: Don't crash if the driver sees an unused input file when running as
'clang-cpp'.
For now, the test uses "REQUIRES: shell" to determine if the host system
supports "ln -s", which it uses to create a 'clang-cpp' symlink. This is a bit
hacky and should likely be directly supported by lit.cfg.
Richard Smith [Mon, 6 Aug 2012 03:25:17 +0000 (03:25 +0000)]
PR13499: Don't try to check whether 'override' has been validly applied until
we know whether the function is virtual. But check it as soon as we do know;
in some cases we don't need to wait for an instantiation.
Jordan Rose [Sat, 4 Aug 2012 00:25:30 +0000 (00:25 +0000)]
[analyzer] Don't assume values bound to references are automatically non-null.
While there is no such thing as a "null reference" in the C++ standard,
many implementations of references (including Clang's) do not actually
check that the location bound to them is non-null. Thus unlike a regular
null dereference, this will not cause a problem at runtime until the
reference is actually used. In order to catch these cases, we need to not
prune out paths on which the input pointer is null.
Jordan Rose [Fri, 3 Aug 2012 23:09:01 +0000 (23:09 +0000)]
[analyzer] When a symbol is null, we should track its constraints.
Because of this, we would previously emit NO path notes when a parameter
is constrained to null (because there are no stores). Now we show where we
made the assumption, which is much more useful.
Jordan Rose [Fri, 3 Aug 2012 23:08:42 +0000 (23:08 +0000)]
[analyzer] FindLastStoreBRVisitor was not actually finding stores.
The visitor walks back through the ExplodedGraph as expected, but
it wasn't actually keeping track of when a value was assigned. This
meant that it only worked when the value was assigned when the variable
was defined.
Tests in the next commit (dependent on another change).
Anna Zaks [Fri, 3 Aug 2012 21:43:37 +0000 (21:43 +0000)]
[analyzer] ObjC Inlining: Start tracking dynamic type info in the GDM
In the following code, find the type of the symbolic receiver by
following it and updating the dynamic type info in the state when we
cast the symbol from id to MyClass *.
Anna Zaks [Fri, 3 Aug 2012 18:30:18 +0000 (18:30 +0000)]
[analyzer] Malloc: track non-allocated but freed memory
There is no reason why we should not track the memory which was not
allocated in the current function, but was freed there. This would
allow to catch more use-after-free and double free with no/limited IPA.
Also fix a realloc issue which surfaced as the result of this patch.
Benjamin Kramer [Fri, 3 Aug 2012 08:39:58 +0000 (08:39 +0000)]
Fix failed to generate vtables in certain cases.
By C++ standard, the vtable should be generated if the first non-inline
virtual function is defined in the TU. Current version of clang doesn't
generate vtable if the first virtual function is defaulted, because the
key function is regarded as the defaulted function.
Anna Zaks [Thu, 2 Aug 2012 23:41:05 +0000 (23:41 +0000)]
[analyzer] Solve another source of non-determinism in the diagnostic
engine.
The code that was supposed to split the tie in a deterministic way is
not deterministic. Most likely one of the profile methods uses a
pointer. After this change we do finally get the consistent diagnostic
output. Testing this requires running the analyzer on large code bases
and diffing the results.
projects / clang / log