Commit Graph

109692 Commits

Author SHA1 Message Date
Francois Pichet b23dc0950b In Microsoft mode, if we are within a templated function and we can't resolve Identifier during BuildCXXNestedNameSpecifier, then extend the SS with Identifier. This will have the effect of resolving Identifier during template instantiation. The goal is to be able to resolve a function call whose nested-name-specifier is located inside a dependent base class.
class C {
public:
    static void foo2() {  }
};

template <class T> class A {
public:
   typedef C D;
};

template <class T> class B : public A<T> {
public:
  void foo() { D::foo2(); }
};

Note that this won't work if the NestedNameSpecifier refers to a type.
This fixes 1 error when parsing the MSVC 2010 standard headers file with clang.

llvm-svn: 136203
2011-07-27 01:05:24 +00:00
Eli Friedman 8b5277c6cf Minor simplification.
llvm-svn: 136202
2011-07-27 01:02:25 +00:00
Bruno Cardoso Lopes f9324f4f6b Move some code around to open opportunity for more shuffle matching
llvm-svn: 136201
2011-07-27 00:56:37 +00:00
Bruno Cardoso Lopes 27a30a7792 The vpermilps and vpermilpd have different behaviour regarding the
usage of the shuffle bitmask. Both work in 128-bit lanes without
crossing, but in the former the mask of the high part is the same
used by the low part while in the later both lanes have independent
masks. Handle this properly and and add support for vpermilpd.

llvm-svn: 136200
2011-07-27 00:56:34 +00:00
Bruno Cardoso Lopes db5fb91491 Remove more dead code!
llvm-svn: 136199
2011-07-27 00:56:27 +00:00
Eli Friedman ae8161e774 Fix AliasSetTracker so that it doesn't make any assumptions about instructions it doesn't know about (like the atomic instructions I'm adding).
llvm-svn: 136198
2011-07-27 00:46:46 +00:00
Evan Cheng 481ebb0133 Support .code32 and .code64 in X86 assembler.
llvm-svn: 136197
2011-07-27 00:38:12 +00:00
Devang Patel f098ce2757 It is quiet possible that inlined function body is split into multiple chunks of consequtive instructions. But, there is not any way to describe this in .debug_inline accelerator table used by gdb. However, describe non contiguous ranges of inlined function body appropriately using AT_range of DW_TAG_inlined_subroutine debug info entry.
llvm-svn: 136196
2011-07-27 00:34:13 +00:00
Johnny Chen 2af257cd0d Add deprecation docstrings for some APIs which take an SBFrame - SBValues know their own frames.
llvm-svn: 136194
2011-07-27 00:08:59 +00:00
Eric Christopher d23b96f923 Remove these two directories. The tests can be ported to dragonegg if
they're still wanted.

llvm-svn: 136193
2011-07-27 00:07:56 +00:00
Eric Christopher 12ff9528b2 Handle different sized wchar_t for windows.
llvm-svn: 136192
2011-07-26 23:56:49 +00:00
Eric Christopher 4481732015 Remove test/FrontendC, almost all of the tests have been migrated
to clang now, the rest are in process (6) or have been deleted.

llvm-svn: 136191
2011-07-26 23:49:39 +00:00
Ted Kremenek fbcce6fba3 clang_getCXTUResourceUsage: report memory used by HeaderSearch.
This required converting the StringMaps to use a BumpPtrAllocator.  I measured the
compile time and saw no observable regression.

llvm-svn: 136190
2011-07-26 23:46:11 +00:00
Ted Kremenek 120992ad81 clang_getCXTUResourceUsage: Report memory used by data structures in SourceManager.
llvm-svn: 136189
2011-07-26 23:46:06 +00:00
Eric Christopher 99c3638935 Use the correct definition for memset.
llvm-svn: 136188
2011-07-26 23:44:22 +00:00
Johnny Chen 33c0c77073 Fix indentation and add a docstring for the ctor.
llvm-svn: 136187
2011-07-26 23:42:01 +00:00
Jakob Stoklund Olesen dab4b9a4b2 Add support for multi-way live range splitting.
When splitting global live ranges, it is now possible to split for
multiple destination intervals at once. Previously, we only had the main
and stack intervals.

Each edge bundle is assigned to a split candidate, and splitAroundRegion
will insert copies between the candidate intervals and the stack
interval as needed.

The multi-way splitting is used to split around compact regions when
enabled with -compact-regions. The best candidate register still gets
all the bundles it wants, but everything outside the main interval is
first split around compact regions before we create single-block
intervals.

Compact region splitting still causes some regressions, so it is not
enabled by default.

llvm-svn: 136186
2011-07-26 23:41:46 +00:00
Eric Christopher fde54ab8b5 Make this darwin only to simplify the padding calculations per arch.
llvm-svn: 136185
2011-07-26 23:39:23 +00:00
Johnny Chen 94b431ab63 Add skip test for clang, which has insufficient debug info for call site in main().
llvm-svn: 136184
2011-07-26 23:35:38 +00:00
Eli Friedman 1b71a22b28 Re-fix r136172 so it isn't an error; apparently, some people are fond of their undefined behavior.
llvm-svn: 136183
2011-07-26 23:27:24 +00:00
Eric Christopher 8f95f425aa This appears to be passing on all hosts.
llvm-svn: 136182
2011-07-26 23:26:00 +00:00
Jeffrey Yasskin a6667816d5 This patch implements as much of the narrowing conversion error specified by
[dcl.init.list] as is possible without generalized initializer lists or full
constant expression support, and adds a c++0x-compat warning in C++98 mode.

The FixIt currently uses a typedef's basename without qualification, which is
likely to be incorrect on some code.  If it's incorrect on too much code, we
should write a function to get the string that refers to a type from a
particular context.

The warning is currently off by default. I'll fix LLVM and clang before turning
it on.

llvm-svn: 136181
2011-07-26 23:20:30 +00:00
Douglas Gregor 8771796493 Disable the optimization that skips emission of complete, non-virtual
destructors of abstract classes. It's undefined behavior to actually
call the destructor (e.g., via delete), but the presence of code that
calls this destructor doesn't make the program
ill-formed. Fixes <rdar://problem/9819242>.

llvm-svn: 136180
2011-07-26 23:18:30 +00:00
Eric Christopher ebc113d1d3 Try a little hack to fix the memset duplication on windows.
llvm-svn: 136179
2011-07-26 23:18:10 +00:00
Jakob Stoklund Olesen b1459dbc25 Print out the MBB live-in registers.
llvm-svn: 136178
2011-07-26 23:12:08 +00:00
Eric Christopher 9688a17bff Try harder to fix these for windows.
llvm-svn: 136177
2011-07-26 23:12:06 +00:00
Jakob Stoklund Olesen c3bcb02154 Eliminate copies of undefined values during coalescing.
These copies would coalesce easily, but the resulting value would be
defined by a deleted instruction. Now we also remove the undefined value
number from the destination register.

This fixes PR10503.

llvm-svn: 136174
2011-07-26 23:00:24 +00:00
Eric Christopher 28f8189866 Generalize for various build bots.
llvm-svn: 136173
2011-07-26 22:52:35 +00:00
Eli Friedman 1260f59f5e Diagnose trying to delete a pointer to an abstract class with a non-virtual destructor. PR10504.
I'm not completely sure the standard allows us to reject this, but if it doesn't, it should. :)

llvm-svn: 136172
2011-07-26 22:50:18 +00:00
Benjamin Kramer a79c1e0589 Update test.
llvm-svn: 136170
2011-07-26 22:45:39 +00:00
Eric Christopher 19520147fc Rewrite match line to be friendlier to misc buildbots.
llvm-svn: 136169
2011-07-26 22:44:31 +00:00
Eric Christopher cf2ea54160 Rewrite matching line to be friendlier to misc buildbots.
llvm-svn: 136168
2011-07-26 22:43:37 +00:00
Benjamin Kramer 124ac2b997 Add a neat little two's complement hack for x86.
On x86 we can't encode an immediate LHS of a sub directly. If the RHS comes from a XOR with a constant we can
fold the negation into the xor and add one to the immediate of the sub. Then we can turn the sub into an add,
which can be commuted and encoded efficiently.

This code is generated for __builtin_clz and friends.

llvm-svn: 136167
2011-07-26 22:42:13 +00:00
Eric Christopher f883941903 Attempt to rewrite the matching for this test to pacify the windows
build bots.

llvm-svn: 136166
2011-07-26 22:42:01 +00:00
Anna Zaks 6868b35f01 Remove unused diagnostic definition.
llvm-svn: 136165
2011-07-26 22:34:49 +00:00
Eric Christopher 970fa44d70 Rework the match here to silence release mode windows buildbots.
llvm-svn: 136164
2011-07-26 22:32:44 +00:00
Eric Christopher 738c7ef5bb Attempt to generalize this test for release mode buildbots.
llvm-svn: 136163
2011-07-26 22:31:33 +00:00
Eric Christopher 4a8bc43083 Try to generalize the match to quiet the windows build bot.
llvm-svn: 136162
2011-07-26 22:27:01 +00:00
Eli Friedman ae4280f721 A couple minor issues with Sema for delete:
1. Attempting to delete an expression of incomplete class type should be an error, not a warning.

2. If someone tries to delete a pointer to an incomplete class type, make sure we actually emit
the delete expression after we warn.

llvm-svn: 136161
2011-07-26 22:25:31 +00:00
Eric Christopher aee57b4948 Use unsigned long long for uint64_t. Fixes part of the windows buildbot.
llvm-svn: 136160
2011-07-26 22:19:01 +00:00
Eric Christopher 85e5156598 Migrate most of the rest of test/FrontendC from llvm and migrate
most of them to FileCheck.

llvm-svn: 136159
2011-07-26 22:17:02 +00:00
Eric Christopher 2dfed48cae Adjust check for release mode.
llvm-svn: 136158
2011-07-26 22:07:13 +00:00
Bruno Cardoso Lopes f8fe47bd2b Recognize unpckh* masks and match 256-bit versions. The new versions are
different from the previous 128-bit because they work in lanes.
Update a few comments and add testcases

llvm-svn: 136157
2011-07-26 22:03:40 +00:00
Dan Gohman 9eb62cd159 Delete unnecessarily cautious LastCALLSEQ code.
llvm-svn: 136156
2011-07-26 22:00:59 +00:00
Argyrios Kyrtzidis e3be9795c8 In ARC we emit an error when compiling:
@interface Foo : NSObject
@property (readonly) id myProp;
@end

@implementation Foo
@synthesize myProp;
@end

t.m:9:13: error: ARC forbids synthesizing a property of an Objective-C object with unspecified storage attribute
@synthesize myProp;
            ^

which is fine, we want the ownership of the synthesized ivar to be explicit. But we should _not_ emit an error
for the following cases, because we can get the ownership either from the declared ivar or from the property type:

@interface Foo : NSObject {
	__weak id _myProp1;
	id myProp2;
}
@property (readonly) id myProp1;
@property (readonly) id myProp2;
@property (readonly) __strong id myProp3;
@end

@implementation Foo
@synthesize myProp1 = _myProp1;
@synthesize myProp2;
@synthesize myProp3;
@end

rdar://9844006.

llvm-svn: 136155
2011-07-26 21:48:26 +00:00
Jim Grosbach a5f7a8cfde ARM rot_imm printing adjustment.
Allow the rot_imm operand to be optional. This sets the stage for refactoring
away the "rr" versions from the multiclasses and replacing them with Pat<>s.

llvm-svn: 136154
2011-07-26 21:44:37 +00:00
Eric Christopher e70ea8b806 Migrate:
CodeGen/2003-08-21-WideString.c
       CodeGen/2003-10-02-UnionLValueError.c
       CodeGen/2004-02-20-Builtins.c
       CodeGen/2008-01-04-WideBitfield.c
       CodeGen/2002-07-14-MiscTests3.c
       CodeGen/2005-04-09-ComplexOps.c
       CodeGen/2008-12-23-AsmIntPointerTie.c
       CodeGen/2005-07-20-SqrtNoErrno.c
       CodeGen/2005-01-02-VAArgError-ICE.c
       CodeGen/2004-06-17-UnorderedCompares.c
       CodeGen/2002-06-25-FWriteInterfaceFailure.c
       CodeGen/2002-02-18-64bitConstant.c
       CodeGen/2002-05-24-Alloca.c
       CodeGen/2006-01-13-Includes.c
       CodeGen/2007-09-27-ComplexIntCompare.c
       CodeGen/2004-02-13-IllegalVararg.c
       CodeGen/2007-09-12-PragmaPack.c
       CodeGen/2002-08-02-UnionTest.c

from test/FrontendC with changes to remove header file includes.

llvm-svn: 136153
2011-07-26 21:42:32 +00:00
Jim Grosbach d2659138da ARM cleanup of rot_imm encoding.
Start of cleaning this up a bit. First step is to remove the encoder hook by
storing the operand as the bits it'll actually encode to so it can just be
directly used. Map it to the assembly source values 8/16/24 when we print it.

llvm-svn: 136152
2011-07-26 21:28:43 +00:00
Ted Kremenek 182543aba2 Report more memory using in Preprocessor::getTotalMemory() and PreprocessingRecord::getTotalMemory().
Most of the memory was already reported; but now we report more memory from side data structures.

Fixes <rdar://problem/9379717>.

llvm-svn: 136150
2011-07-26 21:17:24 +00:00
Eli Friedman 93dc04d5ca Prevent x86-specific DAGCombine from creating nodes with illegal type (which could not be selected). Fixes a minor isel issue that was breaking the testcase from r136130.
llvm-svn: 136148
2011-07-26 21:02:58 +00:00