Add a FileCharacteristic parameter to SourceManager::createFileIDForMemBuffer
for completeness and use it in CompilerInstance::InitializeSourceManager if
the input is a memory buffer.
Andy Gibbs [Fri, 9 Nov 2012 13:24:30 +0000 (13:24 +0000)]
Improved support for removing the comma preceding __VA_ARGS__ where __VA_ARGS__
is empty in a variadic macro expansion. This fixes a divergence in support for
the ", ## __VA_ARGS__" GCC extension which differed in behaviour when in strict
C99 mode (note: there is no change in behaviour has been made in the gnu99 mode
that clang uses by default). In addition, there is improved support for the
Microsoft alternative extension ", __VA_ARGS__".
Fix a bug I found while preparing my devmtg talk: When passing NULL to a
function that takes a const Foo&, where Foo is convertible from a large number
of pointer types, we print ALL the overloads, no matter the setting of
-fshow-overloads.
There is potential follow-on work in unifying the "print candidates, but not
too many" logic between OverloadCandidateSet::NoteCandidates and
ImplicitConversionSequence::DiagnoseAmbiguousConversion.
Benjamin Kramer [Thu, 8 Nov 2012 12:59:15 +0000 (12:59 +0000)]
Set TLSSupported to false on Haiku.
Haiku does not support this (yet). Leaving it set to true leads to
configure scripts detecting __thread being available and Clang emitting
code for it, resulting in binaries the runtime_loader will refuse to
load.
Richard Smith [Wed, 7 Nov 2012 23:56:21 +0000 (23:56 +0000)]
When deciding whether to convert an array construction loop into a memcpy, look
at whether the *selected* constructor would be trivial rather than considering
whether the array's element type has *any* non-trivial constructors of the
relevant kind.
Dmitri Gribenko [Wed, 7 Nov 2012 23:52:25 +0000 (23:52 +0000)]
Force C89 for c-index-test.c
MSVC supports only C89, so it is important to keep c-index-test.c
buildable with C89. However, Clang defaults to C99, so while building
Clang with Clang one can introduce C99 constructs into c-index-test.c
without noticing.
Thanks to Nakamura Takumi for helping with MSVC bits.
Richard Trieu [Wed, 7 Nov 2012 21:17:13 +0000 (21:17 +0000)]
When template deduction fails on a derived class, try a template deduction on
the base class. If the base class deduction succeeds, use those results. If
it fails, keep using the results from the derived class template deduction.
This prevents an assertion later where the type of deduction failure doesn't
match up with the template deduction info.
David Chisnall [Wed, 7 Nov 2012 16:50:40 +0000 (16:50 +0000)]
Fix the Objective-C exception rethrow from cleanups (GNU runtimes). Note that
a bug in the inliner still causes the wrong thing to happen at -O2 and above
(PR14116).
Richard Smith [Wed, 7 Nov 2012 01:14:25 +0000 (01:14 +0000)]
PR11851 (and duplicates): Whenever a constexpr function is referenced,
instantiate it if it can be instantiated and implicitly define it if it can be
implicitly defined. This matches g++'s approach. Remove some cases from
SemaOverload which were marking functions as referenced when just planning how
overload resolution would proceed; such cases are not actually references.
David Blaikie [Wed, 7 Nov 2012 00:12:38 +0000 (00:12 +0000)]
PR13552: Fix the end location of a CXXNewExpr.
Spent longer than reasonable looking for a nice way to test this & decided to
give up for now. Open to suggestions/requests. Richard Smith suggested adding
something to ASTMatchers but it wasn't readily apparent how to test this with
that.
Eli Friedman [Tue, 6 Nov 2012 23:40:48 +0000 (23:40 +0000)]
Put something sane in the DWARF offset field for bitfield ObjC ivars.
This is useful because unnamed bitfields can have effects on the
offsets which are not otherwise reflected in the DWARF information.
Implement codegen for init_priority attribute properly - make sure it
works between the modules.
No functionality change on Darwin/Windows.
This fixes PR11480.
Douglas Gregor [Tue, 6 Nov 2012 19:39:40 +0000 (19:39 +0000)]
Introduce inferred framework modules into the module map file,
allowing a module map to be placed one level above the '.framework'
directories to specify that all .frameworks within that directory can
be inferred as framework modules. One can also specifically exclude
frameworks known not to work.
This makes explicit (and more restricted) behavior modules have had
"forever", where *any* .framework was assumed to be able to be built
as a module. That's not necessarily true, so we white-list directories
(with exclusions) when those directories have been audited.
Michael Han [Tue, 6 Nov 2012 19:34:54 +0000 (19:34 +0000)]
Teach Clang parser to reject C++11 attributes that appertain to declaration specifiers.
We don't support any C++11 attributes that appertain to declaration specifiers so reject
the attributes in parser until we support them; this also conforms to what g++ 4.8 is doing.
[c-index-test] When building with BUILD_CLANG_ONLY=YES, include c-index-test.
It is part of libclang and has other uses besides running the clang tests.
Ted Kremenek [Tue, 6 Nov 2012 06:36:45 +0000 (06:36 +0000)]
Remove completely irrelevant code from USR generation.
Thanks to Richard Smith for pointing this out. This code stopped
serving its purpose during r103212 in a refactoring. My initial
fix was to add back the logic to abort the USR generation for
InternalLinkage, but enough tests broke suspiciously that I fear
that USR generation for cursors with InternalLinkage is now expected
by some clients (where it wasn't the case when the refactoring
took place). I don't own this code anymore and have not looked
at it for some time, but clearly this code is dead and can be removed
pending further review on the proper logic here.
Richard Smith [Tue, 6 Nov 2012 01:12:02 +0000 (01:12 +0000)]
Per discussion on cfe-commits, treat -faddress-sanitizer,
-fno-address-sanitizer, -fthread-sanitizer, -fno-thread-sanitizer, and
-fcatch-undefined-behavior as deprecated: produce a warning if they are used
pointing to the corresponding -fsanitize= option. In passing add the missing
'-' to some diagnostics.
[PCH] Write out the ClassTemplateDecl::Common::InjectedClassNameType type
reference instead of relying on computing it.
In general, if storage is no issue, it is preferable to deserialize info from
the PCH instead of trying to recompute it after the PCH was loaded.
The incentive to change this now was due to r155303 changing how friend template
classes in dependent contexts are handled; such classes can now be chained to
a previous template class but the computed InjectedClassNameType may be different
due to the extra template parameters from the dependent context.
The new handling requires more investigation but, in the meantime, writing out
InjectedClassNameType fixes PCH issue in rdar://12627738.
Douglas Gregor [Mon, 5 Nov 2012 23:58:27 +0000 (23:58 +0000)]
Have the parser initialize Sema before it consumes the first
token. This is important because the first token could actually be
after an #include that triggers a module import, which might use
either Sema or the AST consumer before it would have been initialized.
Daniel Dunbar [Mon, 5 Nov 2012 22:53:33 +0000 (22:53 +0000)]
Frontend: Add support for reading named pipes as the main file.
- The whole {File,Source}Manager is built around wanting to pre-determine the
size of files, so we can't fit this in naturally. Instead, we handle it like
we do STDIN, where we just replace the main file contents upfront.
Manman Ren [Mon, 5 Nov 2012 22:42:46 +0000 (22:42 +0000)]
ARM byval: when type alignment is bigger than ABI alignment, we can't guarantee
the type alignment of the byval argument. This patch will disable byval in this case,
it also increases the size threshold for turning on byval.
Richard Smith [Mon, 5 Nov 2012 22:21:05 +0000 (22:21 +0000)]
Use the individual -fsanitize=<...> arguments to control which of the UBSan
checks to enable. Remove frontend support for -fcatch-undefined-behavior,
-faddress-sanitizer and -fthread-sanitizer now that they don't do anything.
Richard Smith [Mon, 5 Nov 2012 22:04:41 +0000 (22:04 +0000)]
Add -fsanitize=<sanitizers> argument to driver and frontend, and add
-fno-sanitize=<sanitizers> argument to driver. These allow ASan, TSan, and the
various UBSan checks to be enabled and disabled separately. Right now, the
different modes can't be combined, but the intention is that combining UBSan
and the other sanitizers will be permitted in the near future.
Currently, the UBSan checks will all be enabled if any of them is; that will be
fixed by the next patch.
Douglas Gregor [Mon, 5 Nov 2012 20:11:10 +0000 (20:11 +0000)]
Make cpuid.h actually work with -std=c99 <rdar://problem/12552716>.
While we're here, extend the module map to cover most of the
newly-added instrinsic headers. Only wmmintrin.h is missing, because
it needs to be split into AES/PCLMUL subheaders (as a separate commit).
Douglas Gregor [Mon, 5 Nov 2012 19:45:09 +0000 (19:45 +0000)]
Hash the various compiler version, target, preprocessor, and
header-search options into the module hash. We're just using
ADT/Hashing.hpp for this, which isn't as cryptographically strong as
I'd like, but it'll do. If someone contributes (say) and MD4
implementation, we'd happily switch to that.
Ulrich Weigand [Mon, 5 Nov 2012 19:13:42 +0000 (19:13 +0000)]
On PowerPC64, integer arguments and return values need to be sign- or
zero-extended to 64 bits. This information is currently provided to
the back end by setting "signext" or "zeroext" attributes. However,
this is done only for integer types *smaller* than i32, not for i32
itself. This causes clang to generate code violating the ABI, which
results in a failure of the tramp3d-v4 test case (due to calling a
system library routine without ABI-required extension).
This patch implements custom versions of classifyArgumentType and
classifyReturnType for PPC64_SVR4_ABIInfo, which are the same as the
default versions except that they also classify "int" and "unsigned int"
as types needing extending. This fixed tramp3d-v4 on PowerPC64.