James Dennett [Fri, 9 Aug 2013 23:08:25 +0000 (23:08 +0000)]
Expose LambdaIntroducer::DefaultLoc in the AST's LambdaExpr.
Summary:
Source-centric tools need access to the location of a C++11
lambda expression's capture-default ('&' or '=') when it's present.
It's possible for them to find it by re-lexing and re-implementing
rules that Clang's parser has already applied, but the cost of storing
the SourceLocation and making it available to them is 32 bits per
LambdaExpr (a small delta, proportionally), and the simplification in
client code is significant.
Richard Trieu [Fri, 9 Aug 2013 21:42:32 +0000 (21:42 +0000)]
Fix for PR16570: when comparing two function pointers, discard qualifiers when
comparing non-reference function parameters. The qualifiers don't matter for
comparisons.
This is a re-commit of r187769, which was accidentially reverted in r187770,
with a simplification at the suggestion of Eli Friedman.
Pavel Labath [Fri, 9 Aug 2013 07:46:29 +0000 (07:46 +0000)]
[analyzer] Enable usage of temporaries in InitListExprs
Summary:
ExprEngine had code which specificaly disabled using CXXTempObjectRegions in
InitListExprs. This was a hack put in r168757 to silence a false positive.
The underlying problem seems to have been fixed in the mean time, as removing
this code doesn't seem to break anything. Therefore I propose to remove it and
solve PR16629 in the process.
Richard Smith [Fri, 9 Aug 2013 04:35:01 +0000 (04:35 +0000)]
Implement [class.friend]p11's special name lookup rules for friend declarations
of local classes. We were previously handling this by performing qualified
lookup within a function declaration(!!); replace it with the proper scope
lookup.
Hans Wennborg [Fri, 9 Aug 2013 00:32:23 +0000 (00:32 +0000)]
clang-cl: Support /showIncludes
This option prints information about #included files to stderr. Clang could
already do it, this patch just teaches the existing code about the /showIncludes
style and adds the flag.
Reid Kleckner [Thu, 8 Aug 2013 19:33:10 +0000 (19:33 +0000)]
Revert "clang-cl: Hook up /Za to prevent adding oldnames.lib dependency"
This reverts commit r187991 and adjusts the comment. /Za is much more
involved, and we don't want to give anyone the impression we actually
support it.
The only useful loop unrolling flag to give realistically is
'-fno-unroll-loops'. The option to the backend is even called
'DisableUnrollLoops'. This is precisely the form that Clang *didn't*
support. We didn't recognize the flag, we didn't pass it to the CC1
layer, and even if we did we wouldn't use it. Clang only inspected the
positive form of the flag, and only did so to enable loop unrolling when
the optimization level wasn't high enough. This only occurs for an
optimization level that even has a chance of running the loop unroller
when optimizing for size.
This commit wires up the 'no' variant, and switches the code to actually
follow the standard flag pattern of using the last flag and allowing
a flag in either direction to override the default.
I think this is still wrong. I don't know why we disable the loop
unroller entirely *from Clang* when optimizing for size, as the loop
unrolling pass *already has special logic* for the case where the
function is attributed as optimized for size! We should really be
trusting that. Maybe in a follow-up patch, I don't really want to change
behavior here.
Summary:
UBSan was checking for alignment of the derived class on the pointer to
the base class, before converting. With some class hierarchies, this could
generate false positives.
DataFlowSanitizer is a generalised dynamic data flow analysis.
Unlike other Sanitizer tools, this tool is not designed to detect a
specific class of bugs on its own. Instead, it provides a generic
dynamic data flow analysis framework to be used by clients to help
detect application-specific issues within their own code.
[PCH] Fix a PCH serialization crash, with invalid code related to forward enum references.
The problem was that an enum without closing semicolon could be associated as a forward enum
in an erroneous declaration, leading to the identifier being associated with the enum decl but
without a declaration actually referencing it.
This resulted in not having it serialized before serializing the identifier that is associated with.
Also prevent the ASTUnit from querying the serialized DeclID for an invalid top-level decl; it may not
have been serialized.
Manuel Klimek [Wed, 7 Aug 2013 19:20:45 +0000 (19:20 +0000)]
Fixes a couple of bugs with the Allman brace breaking.
In particular, left braces after an enum declaration now occur on their
own line. Further, when short ifs/whiles are allowed these no longer
cause the left brace to be on the same line as the if/while when a
brace is included.
This field is just IsDefaulted && !IsDeleted; in all places it's used,
a simple check for isDefaulted() is superior anyway, and we were forgetting
to set it in a few cases.
Also eliminate CXXDestructorDecl::IsImplicitlyDefined, for the same reasons.
Daniel Jasper [Wed, 7 Aug 2013 05:34:02 +0000 (05:34 +0000)]
clang-format: Improve formatting of builder-type calls.
This removes a formatting choice that was added at one point, but is
not generally liked by users. Specifically, in builder-type calls, do
(easily) break if the object before the ./-> is either a field or a
parameter-less function call. I.e., don't break after "aa.aa.aa" or
"aa.aa.aa()". In general, these sequences in builder-type calls are
seen as a single entity and thus breaking them up is a bad idea.
ObjectiveC migration: tweak setting of lifetime attribute
on @property migration. Don't set unsafe_unretained
on non-object properties. Set 'retain' on strong
properties. Makecertain properties with specific
names unsafe_unretained as well.
Larisse Voufo [Tue, 6 Aug 2013 05:49:26 +0000 (05:49 +0000)]
Removed hack that was used to properly restore the nested name specifier of qualified variable template ids. It turns out that the current implementation was just not logical setup for it. This commit has made it so.
Richard Trieu [Tue, 6 Aug 2013 03:44:10 +0000 (03:44 +0000)]
Fix for PR16570: when comparing two function pointers, discard qualifiers when
comparing non-reference function parameters. The qualifiers don't matter for
comparisons.
Larisse Voufo [Tue, 6 Aug 2013 01:03:05 +0000 (01:03 +0000)]
Started implementing variable templates. Top level declarations should be fully supported, up to some limitations documented as FIXMEs or TODO. Static data member templates work very partially. Static data member templates of class templates need particular attention...
David Majnemer [Mon, 5 Aug 2013 22:43:06 +0000 (22:43 +0000)]
[ms-cxxabi] Properly mangle member pointers
There were three things missing from the original implementation:
- We would omit the 'E' qualifier for members int 64-bit mode.
- We would not exmaine the qualifiers in 'IsMember' mode.
- We didn't generate the correct backref to the base class.
Hans Wennborg [Mon, 5 Aug 2013 20:14:43 +0000 (20:14 +0000)]
lit.cfg: better check for MSYS
When running the tests under Cygwin using non-Cygwin python,
the platform would be Windows and there would be bash on the path,
so this check for MSYS would not work correctly.
Richard Smith [Mon, 5 Aug 2013 18:49:43 +0000 (18:49 +0000)]
Implement C++'s restrictions on the type of an expression passed to a vararg
function: it can't be 'void' and it can't be an initializer list. We give a
hard error for these rather than treating them as undefined behavior (we can
and probably should do the same for non-POD types in C++11, but as of this
change we don't).
Slightly rework the checking of variadic arguments in a function with a format
attribute to ensure that certain kinds of format string problem (non-literal
string, too many/too few arguments, ...) don't suppress this error.
Craig Topper [Mon, 5 Aug 2013 06:17:21 +0000 (06:17 +0000)]
Use a shuffle with undef elements instead of inserting 0s in the 128-bit to 256-bit casting intrinsics to improve performance. Thanks to Katya Romanova for identifying this issue.
David Majnemer [Mon, 5 Aug 2013 04:53:41 +0000 (04:53 +0000)]
Sema: Don't assume a nested name specifier holds a type
Sema::PerformObjectMemberConversion assumed that the Qualifier it was
given holds a type. However, the specifier could hold just a namespace.
In this case, we should ignore the qualifier and not attempt to cast to
it.
Hans Wennborg [Fri, 2 Aug 2013 22:24:50 +0000 (22:24 +0000)]
Re-commit r187637: "clang-cl: add more options"
> This adds a bunch of options to clang-cl. Notably, this includes
> all the options that get passed when doing a default build of a
> command-line project with msbuild.exe in Debug and Release modes,
> and I believe all flags from Reid's original patch.
The original commit was reverted in r187640 after it broke the Mac build.
This should now be fixed, by Clang r187668, LLVM r187675, and putting
a -- before %s in the test.
Manuel Klimek [Fri, 2 Aug 2013 21:24:09 +0000 (21:24 +0000)]
Fix crash when encountering alias templates in isDerivedFrom matches.
- pull out function to drill to the CXXRecordDecl from the type,
to allow recursive resolution
- make the analysis more robust by rather skipping values we don't
understand
projects / clang / log