Richard Smith [Thu, 13 Sep 2012 21:18:54 +0000 (21:18 +0000)]
Implement C++11 [conv.prom]p4: an enumeration with a fixed underlying type has
integral promotions to both its underlying type and to its underlying type's
promoted type. This matters now that boolean conversions aren't permitted in
converted constant expressions (a la DR1407): an enumerator with a fixed
underlying type of bool still can be.
Douglas Gregor [Thu, 13 Sep 2012 21:01:57 +0000 (21:01 +0000)]
Don't perform template argument deduction against invalid templates;
it's likely to lead to a crash later on. Fixes PR12933 /
<rdar://problem/11525335>.
Dmitri Gribenko [Thu, 13 Sep 2012 20:36:01 +0000 (20:36 +0000)]
Comment parsing: handle \deprecated command. It is a block command, but it
should be fine to use it without further explanations in the attached
paragraph, so the warning about empty paragraph was turned off for it.
Douglas Gregor [Thu, 13 Sep 2012 20:16:20 +0000 (20:16 +0000)]
Promote the warning about extra qualification on a declaration from a
warning to an error. C++ bans it, and both GCC and EDG diagnose it as
an error. Microsoft allows it, so we still warn in Microsoft
mode. Fixes <rdar://problem/11135644>.
Richard Smith [Thu, 13 Sep 2012 18:37:50 +0000 (18:37 +0000)]
Make TextDiagnostic more robust against SourceLocations which point into the
middle of UTF-8 characters, and avoid walking to such positions when adjusting
column ranges for display. Fixes a couple of hangs when rendering diagnostics.
When warning about unsafe uses of dispatch_once, specially handle the
crazy case where dispatch_once gets redefined as a macro that calls
_dispatch_once (which calls the real dispatch_once). Users want to
see the warning in their own code.
Conditionally parse documentation comments in system headers by
passing -fretain-comments-from-system-headers. By default, the
compiler no longer parses such documentation comments, as they
can result in a noticeable compile time/PCH slowdown.
I debated whether or not this logic should be sunk into the CFG
itself. It's not clear if we should, as different analyses may
wish to have different policies. We can re-evaluate this in the
future.
Refactor logic in ExprEngine for detecting 'noreturn' methods
in NSException to a helper object in libAnalysis that can also
be used by Sema. Not sure if the predicate name 'isImplicitNoReturn'
is the best one, but we can massage that later.
Chad Rosier [Thu, 13 Sep 2012 00:06:55 +0000 (00:06 +0000)]
[ms-inline asm] Handle the enumeration of input and output expressions in a
more robust way to address a few FIXMEs.
The initial implementation, r163342, built the IR asm string and then tried to
patch things on the fly without enough context. Specifically, it didn't skip
mnemonics nor did it track with assembly instruction an expression was related
to. The new implementation patches the operands and then builds the final
IR string.
Eric Christopher [Wed, 12 Sep 2012 23:36:49 +0000 (23:36 +0000)]
Reapply:
Make clang emit a flag for DW_AT_object_pointer for the artificial
args where it should (implicit first arguments). FileCheck-ize a
test as well and update tests to take into account the object
pointer flag.
Richard Trieu [Wed, 12 Sep 2012 23:13:31 +0000 (23:13 +0000)]
Add two new functions to Qualifiers.
static Qualifiers removeCommonQualifiers(Qualifiers &L, Qualifiers &R)
Removes the common qualifiers from L and R and returns them in a new Qualifier.
This will be used in template diffing.
void removeQualifiers(Qualifiers Q)
Removes the qualifiers in Q from the current qualifier.
This replaces the current underlying implementation of operator- and
operator -= which only performed bit masking.
Jordan Rose [Wed, 12 Sep 2012 21:48:17 +0000 (21:48 +0000)]
[analyzer] Handle when the dynamic type is worse than the static type.
Currently we don't update the dynamic type of a C++ object when it is
cast. This can cause the situation above, where the static type of the
region is now known to be a subclass of the dynamic type.
Once we start updating DynamicTypeInfo in response to the various kinds
of casts in C++, we can re-add this assert to make sure we don't miss
any cases. This work is tracked by <rdar://problem/12287087>.
In -Asserts builds, we will simply not return any runtime definition
when our DynamicTypeInfo is known to be incorrect like this.
Jordan Rose [Wed, 12 Sep 2012 21:48:13 +0000 (21:48 +0000)]
Revert "[analyzer] Use the static type for a virtual call if the dynamic type is worse."
Using the static type may be inconsistent with later calls. We should just
report that there is no inlining definition available if the static type is
better than the dynamic type. See next commit.
Chad Rosier [Wed, 12 Sep 2012 18:14:25 +0000 (18:14 +0000)]
[ms-inline asm] If we have a single asm operand that maps to multiple
MCOperands then iterate over all of then when computing clobbers, inputs and
outputs.
On x86 the 1-to-many mapping is a memory operand that includes a BaseReg(reg),
MemScale(imm), MemIndexReg(reg), an Expr(MCExpr or imm) and a MemSegReg(reg).
Invalid register (Op.getReg() == 0) are not considered when computing clobber.
Teach -Wuninitialized to recognize __attribute__((analyzer_noreturn))
for halting the propagation of uninitialized value tracking along
a path. Unlike __attribute__((noreturn)), this attribute (which
is used by clients of the static analyzer) can be used to annotate
functions that essentially never return, but in rare cares may be
allowed to return for (special) debugging purposes. This attribute
has been shown in reducing false positives in the static analyzer
by pruning false postives, and is equally applicable here.
Handling this attribute in the CFG itself is another option, but
this is not something all clients (e.g., possibly -Wunreachable-code)
would want to see.
Adjust some analyzer tests to place widely shared inputs inside of an
'Inputs' subdirectory.
The general desire has been to have essentially all of the non-test
input files live in such directories, with some exceptions for obvious
and common patterns like 'foo.c' using 'foo.h'.
This came up because our distributed test runner couldn't find some of
the headers, for example with stl.cpp.
No functionality changed, just shuffling around here.
Richard Smith [Wed, 12 Sep 2012 00:56:43 +0000 (00:56 +0000)]
PR13811: Add a FunctionParmPackExpr node to handle references to function
parameter packs where the reference is not being expanded but the pack has
been. Previously, Clang would segfault in such cases.
Eric Christopher [Wed, 12 Sep 2012 00:27:32 +0000 (00:27 +0000)]
Make clang emit a flag for DW_AT_object_pointer for the artificial
args where it should (implicit first arguments). FileCheck-ize a
test as well and update tests to take into account the object
pointer flag.
Jordan Rose [Tue, 11 Sep 2012 20:26:49 +0000 (20:26 +0000)]
Revert "[analyzer] Disable STL inlining. Blocked by PR13724."
While PR13724 is still an issue, it's not actually an issue in the STL.
We can keep this option around in case there turn out to be widespread
false positives due to poor modeling of the C++ standard library functions,
but for now we'd like to get more data.
Dmitri Gribenko [Tue, 11 Sep 2012 19:22:03 +0000 (19:22 +0000)]
Comment parsing: handle non-builtin commands correctly. After semantic
analysis registers a command, it becomes a "known" command for the lexer, since
it has an ID. Having this freedom of choice to register a command is a good
thing since BriefParser does not need this.
But the parser should still invoke the correct semantic analysis method
(actOnUnknownCommand) in this case.
Jordan Rose [Tue, 11 Sep 2012 18:47:13 +0000 (18:47 +0000)]
[analyzer] Use the static type for a virtual call if the dynamic type is worse.
reinterpret_cast does not provide any of the usual type information that
static_cast or dynamic_cast provide -- only the new type. This can get us
in a situation where the dynamic type info for an object is actually a
superclass of the static type, which does not match what CodeGen does at all.
In these cases, just fall back to the static type as the best possible type
for devirtualization.
David Chisnall [Tue, 11 Sep 2012 09:58:54 +0000 (09:58 +0000)]
Select the correct, or, failing that, compatible, dialect when invoked as cc,
c89, c99, and so on. No change to the default dialect when invoked as clang /
clang++.
Eric Christopher [Tue, 11 Sep 2012 01:36:54 +0000 (01:36 +0000)]
The type of the self and cmd variables should be artificial. (Note
that the types aren't artificial the args are, but this is currently
represented by an artificial type.)
Jordan Rose [Tue, 11 Sep 2012 00:31:02 +0000 (00:31 +0000)]
[analyzer] Member function calls that use qualified names are non-virtual.
C++11 [expr.call]p1: ...If the selected function is non-virtual, or if the
id-expression in the class member access expression is a qualified-id,
that function is called. Otherwise, its final overrider in the dynamic type
of the object expression is called.
projects / clang / log